Explore every episode of the podcast Inside AsembleAI: DeepTech, AI & Science

Walter Haydock has spent his career helping data and security leaders deploy AI without fear. A former U.S. Marine officer and founder of StackAware, he joins Sam and Mac to talk about building a governance foundation before the risk outruns you.

Topics covered:

• The biggest mistake in starting AI governance: policies that are unrealistic and unenforceable
• Why your industry’s risk appetite should drive your AI decisions
• Quantitative risk: comparing the cost of a risk with the cost of the control
• People, then process, then tools, and why it applies to companies of every size
• Why risk avoidance is the simplest and cheapest control
• The Marine Corps model: one accountable leader, advised but not vetoed by security, privacy, and legal
• Who owns AI governance: five patterns, and the trade-offs of each
• The EU AI Act delay vs. the U.S. patchwork of state laws
• How NIST AI RMF and ISO/IEC 42001 work together
• “AI canaries” and indirect prompt injection against agents
• Three human-in-the-loop models: default deny, default allow, and post hoc review
• Where a security leader should spend this quarter’s AI budget
• Why running AI on your own hardware isn’t automatically more secure

Guest Bio: Walter Haydock is the founder of StackAware and author of the Deploy Securely newsletter. A former U.S. Marine officer with a Harvard MBA, he helps executives in highly regulated industries navigate AI risk frameworks, compliance, and governance without banning AI or deploying it blindly.

Connect: Find Walter on LinkedIn. If you work in healthcare, financial technology, or B2B software and want help with AI governance, send him a direct message.

Be Brave.

Subscribe to Inside AsembleAI on YouTube, Spotify, Apple Podcasts, Amazon Music, iHeartRadio, and Podbean.

Vanessa Larco has sat on both sides of the table. She built products at Microsoft, Disney, Twilio, and Box, then led investments at NEA, including Mejuri and Kindred, and watched Robinhood scale to its IPO as a board observer. Sam and Mac talk with her about what actually holds up once the pitch deck closes.

Topics covered:

• Why a founder’s roadmap, and what’s missing from it, shows how they make decisions
• Skill, luck, and the dynamics no investor can predict
• Product market pull: users who would be upset if you took the product away
• Why enterprise AI gets the capital, and where consumer AI is still open
• The hybrid go-to-market challenge of selling AI to small businesses
• AI-native companies vs. companies that use AI behind the scenes without selling it
• The “why now” framework behind Mejuri and Kindred
• Lessons from Robinhood’s founders on urgency, obsession, and speed
• Agent traffic vs. human traffic, and why product teams will need agent metrics
• How a venture firm uses AI agents, and why conviction in diligence stays human

Guest Bio: Vanessa Larco is co-founder of Premise Venture Partners and was previously a Partner at NEA, a fund with over $25 billion in AUM, where she led investments in brands like Mejuri and Kindred. Before venture, she built products at Microsoft, Disney, Twilio, and Box. She was a board observer for Robinhood through its IPO, sits on the board of LatinxVC, and has made backing women-led and underrepresented founders a cornerstone of her approach.

Connect: Find Vanessa on LinkedIn, where she says she’s most responsive. She’s also interested in hearing from anyone working on metrics for AI agents.

Rowan Stewart has watched companies rush AI products to market without the governance to back them up, first at BCG and now at Transcend, where the clients include Robinhood, Brex, and Groupon. Sam and Rowan discuss why governance works best as infrastructure from day one, and why agents raise the stakes.

Topics covered:

• "Pilot purgatory": why promising AI proofs of concept die at the edge of production
• The cost of rushing: lost customer trust and bad press matter more than regulatory fines
• Embedded vs. bolted-on governance: a stack of paperwork and a handshake vs. encoded, auditable controls
• Deterministic controls for non-deterministic systems, and why a system prompt isn't a real control
• Human in the loop at scale: a 100-person company with five agents per employee, and how SMBs should adapt
• Why agents shouldn't inherit their owner's permissions, and why MCP tool restrictions aren't sufficient
• Agent observability: knowing what your agents are doing across the enterprise
• How guardrails let teams move faster, and using signals like denial rates to tune policy
• AI governance vs. responsible AI: business outcomes vs. net benefit to humanity
• The one gap to fix first: do you know what agents you even have running?
• Aligning incentives between legal/compliance and product/data science teams

Guest Bio: Rowan Stewart is an AI and Data Safety Product Leader at Transcend, which helps companies like Robinhood, Brex, and Groupon govern and activate their data responsibly. Previously at BCG, Rowan worked with enterprises building AI and now focuses on governance for AI agents.

Connect: Find Rowan Stewart on LinkedIn. Rowan's profile link is in the show notes.

In this episode of Inside AsembleAI, host Mac Goswami sits down with Dr. Latha Karthigaa — Director & Head of AI Governance at the Global AI Certification Council (GAICC) — on why most companies that think their AI is governed are dangerously wrong, and what responsible AI actually looks like.

What's Covered:

"A Policy Is Not Governance" — The biggest myth in enterprise AI. Latha's highway analogy: governance is the guardrail that lets you drive faster, not the thing slowing you down.

52% of Incidents Are Deepfakes — The stats nobody quotes: 2,000+ logged AI incidents, 400+ in 2025, over half deepfake-driven — including the SpaceX/Elon Musk crypto scam that ran live during a real rocket launch.

The $25M Deepfake Call — The engineer at Arup who wired $25 million after a video call with his CFO and colleagues — all of them deepfakes generated from public videos. Why employee training, not just policy, is the real defense.

When Big AI Fails — IBM Watson's oncology tool shut down over biased data; the Dutch childcare-benefits algorithm that falsely accused thousands and brought down a government. Why Latha says the best governance pros are empathetic humans first.

Who Owns the Risk? — "ChatGPT isn't responsible — you are." Why every AI use case needs a human name next to it, and why AI governance jobs are growing 150% year over year.

What ISO/IEC 42001 Really Changes — The one certifiable AI standard, what it signals to customers, and when a startup should (and shouldn't) pursue it.

Key Quote: "Policy is like potential power. Execution is the real power."

Connect with Latha: LinkedIn: https://www.linkedin.com/in/lathakarthigaa/

GAICC: https://www.gaicc.org/ · YouTube: Latha – AI Governance

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Barr Moses — CEO & co-founder of Monte Carlo, creator of the data observability category and now builder of the agent trust platform — about what it actually takes to make AI trustworthy in production.

What's Covered:

"AI Is Wrong in Your Face" — Barr's framing of the trust gap: dashboards were always wrong quietly in the background; AI is wrong out loud, and it'll argue with you. Why trust is the biggest thing standing between pilots and production.

The Four Layers of Agent Failure — Context, performance, behavior, and output. Why all four can look perfect and the agent still fails — and why you have to watch all of them together.

The Flight That Already Left — The airline agent that recommended a flight that departed that morning. The agent was fine; the context was stale. The most surprising failure mode nobody plans for.

Where to Start — Make ONE agent great, not a hundred. And why the hardest first step is simply defining what "good" even looks like.

The Reinforcement Loop — The idea Barr's most excited about: agents that self-identify what went wrong, propose a fix, submit a PR for human approval, and use it as tomorrow's baseline. Agents that rebuild themselves every day — running in production today.

100% AI-First — Why every line of Monte Carlo's code is AI-generated, how it made them 3–5x faster, and Barr's stoplight analogy for where human-in-the-loop is heading.

Key Quote: "Dashboards are wrong in the background. AI is wrong in your face — it'll argue with you."

Connect with Barr: LinkedIn: Barr Moses : https://www.linkedin.com/in/barrmoses/

Monte Carlo: https://www.montecarlo.ai

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

ActualyzeAI came out of stealth just days before this conversation. Sam sits down with Co-Founder and CTO Sean Lynch to unpack what it means to build a "control plane" that sits between every enterprise application and every AI model - governing access, cost, security, and routing in one place.

Topics covered:

• What a control plane for enterprise AI actually does, and why it requires zero code changes to adopt
• Aggregating inference across OpenAI, Anthropic, Google Bedrock, and self-hosted/on-premises models into a single endpoint
• "Virtual models" — purpose-built model configurations that route requests based on task type (coding, reasoning, agentic work)
• Guardrails: automatic detection and redaction of PII, PHI, API keys, and other sensitive data in the inference stream
• Financial operations as the leading driver of adoption — budgetary controls, spend limits, and team-based tracking
• The coming wave of domestic and open-weight small language models, and why that's expanding the market
• Why model-agnostic infrastructure is critical as the foundation model landscape fragments
• How ActualyzeAI's founding team (formerly of Metacloud, acquired by Cisco) shaped their approach
• ActualyzeAI's design partner program for early enterprise customers

Guest Bio: Sean Leach is Co-Founder and CTO of ActualyzeAI , a company building a governance and security control plane for enterprise AI inference. He and much of the founding team previously worked together at Metacloud, an OpenStack-as-a-service company acquired by Cisco.

Connect: Find Sean on LinkedIn, or visit ActualyzeAI's website to learn about their design partner program.

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Oshri Moyal — CTO & co-founder of Atera, the autonomous IT platform whose agent Robin recently ranked #1 across 15 G2 Summer 2026 reports — about what genuinely autonomous IT actually looks like.

What's Covered:

AI That Fixes, Not Just Chats — Why Robin isn't another chatbot. It navigates complex networks, logs into servers, and takes real action — bounded by company policy and approvals. Oshri's example: when users can't reach shared files, Robin hits the domain controller, adds the user to the right group, and maps the drive on their device — end to end.

The Performance Guarantee — Resolve 50% of Tier 1 and complex Tier 2 tickets in 90 days, or fees are waived. Why Atera can stand behind that after two years in production.

Robin as the First Line — How Robin becomes the front door for every request — across Teams, email, Chrome, and ServiceNow — logging everything and closing the loop after approvals.

"80% Was Security" — Becoming the first in IT management to earn ISO/IEC 42001, and how Robin gets elevated permissions only after a manager approves, then hands them back. As Oshri puts it, "80% of the project was about security, privacy, and safety."

What It Changes for Small IT Teams — Why autonomous AI lets a shop "at least double the size of your customers" without adding headcount — enterprise-grade capability without an enterprise team.

Trust at Scale — With 6 million devices connected, why reliability and certification aren't optional.

Key Quote: "With Robin you can at least double the size of your customers, because you can handle twice the amount of tickets — without increasing headcount."

Connect with Oshri: LinkedIn: Oshri Moyal : https://www.linkedin.com/in/oshr1/

Atera: https://www.atera.com/

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Rajan Koo — CTO of DTEX Systems, chartered engineer, and one of the sharpest voices on insider risk — about how AI has completely rewritten the insider-threat playbook.

What's Covered:

WikiLeaks Without a Human — Insider risk was transformed by the 2010 WikiLeaks incident. Raj explains why a recent AI-driven incident showed the same breach can now happen with no humans involved — pushing DTEX into a new category it calls "AI behavior."

"Nobody Was Malicious" — The story that reframes the whole risk: a manufacturing giant's AI agent, blocked from emailing an oversized report, uploaded confidential data to a public drive and shared the link. No malice — enormous risk. Why most insider risk today is negligent, not malicious.

A Teenager Could Run a Nation-State Attack — The North Korean "IT worker" scheme that funded weapons programs can now be replicated by "one person and a team of AI agents." Speed up, skill level down — the perfect storm.

Who Has the Advantage — Why attackers are ahead right now, and how guardrails meant to prevent misuse can block defenders too.

Policy → Behavioral Compliance — Why the age of checklist policies is over, and how DTEX's "agentic defenders" triage risk at machine speed.

Monitoring Without Surveillance — The honest line between protective monitoring and "creepy Big Brother" — and why it all comes down to proportionality and privacy.

Key Quote: "The technical skill level to execute these really complicated insider threat breaches is now really low. A teenager with the right know-how could just go and execute this."

Connect with Raj: LinkedIn: Rajan Koo : https://www.linkedin.com/in/rajan-koo-2a591221/

DTEX: https://www.dtex.ai/

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

Most robotics companies are chasing industrial automation. Mind Children, co-founded by Chris Kudla and Ben Goertzel in 2023, is going after something harder: social robots built for education, healthcare, and hospitality — applications where connection and empathy matter as much as function.

Topics covered:

• Why Mind Children bet on social robotics despite a harder-to-prove ROI than industrial robots
• Cody's modular operating system — and what changes when you swap a standard LLM for SingularityNET's memory-equipped, agentic systems
• The hide-and-seek demo: how Cody reasons in real time and recalls the game a week later
• The emotional intelligence roadmap — teaching Cody to recognize and respond to human distress
• Why Mind Children avoids streaming classroom or hospital video data, and their in-house data approach for regulated environments
• How Chris (product design) and Ben Goertzel (social robotics research) divide responsibilities
• Mind Children's next hardware iteration — designed to be safe enough for a child to hug
• Pilot plans: starting with museums, galleries, and event spaces before schools and healthcare
• How to follow Mind Children's progress and support their crowdfunding campaign on WeFunder

Guest Bio: Chris Kudla is Co-Founder and CEO of Mind Children, a Seattle-based social robotics and AI startup he founded in 2023 alongside Ben Goertzel. Mind Children is building Cody, a social robot for education, healthcare, and hospitality applications.

Connect: Find Mind Children on LinkedIn, at mindchildren.com, or support their campaign at wefunder.com/mindchildrenrobotics.

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Ed Martin — VP of Product Management for AI Strategy at Sophos, with 15+ years in cybersecurity across Dell, SecureWorks, BlueVoyant, and inside Microsoft's Security Deputy CISO office — for a two-in-one conversation on AI-powered defense and what AI is doing to product management itself.

What's Covered:

What AI Actually Does in Defense — Ed's honest cut through the hype: AI is great at reading, good at reasoning, and "moderate to poor at taking action." Where it genuinely helps — summarization, reasoning over big data, reducing analyst toil — and where it's oversold.

The Attacker's Real Edge Is Speed — Why the breadth was always there, but AI lets lower-skill attackers hit harder and faster. Ed's line: "The attacker only has to be right once. We have to be right 100% of the time" — plus the story of a small regional bank suddenly inundated with alerts.

The Cybersecurity Poverty Line — Sophos's mission to protect organizations "at or below the cybersecurity poverty line," and why visibility and hygiene — not hiring — come first.

The PRD Is Dead — Ed's boldest take: the weeks-long product requirements document is gone. His two principal PMs each do the work of a small team, vibe-coding examples live on customer calls. The real skill now is knowing what NOT to build — "it's the scaling that takes all the effort."

Trust & Non-Deterministic AI — Why a single wrong non-deterministic outcome can blow customer trust, and how "circuit breaker" human checkpoints keep AI actions safe.

Building Cyber From Scratch — Inventory first, then controls, then detection and response — and the context/data-categorization problem nearly every organization gets wrong.

Key Quote: "The most important aspect of a product manager isn't the ability to understand what to build. It's your ability to understand what not to build."

Connect with Ed: LinkedIn: Ed Martin : https://www.linkedin.com/in/bigedmartin/

Sophos: https://www.sophos.com/

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

Ben Goertzel coined the term AGI over two decades ago — long before OpenAI, Anthropic, or the current wave of AI labs existed. Sam sits down with him to talk about why he still believes scaling transformers alone won't get the field to true artificial general intelligence, and what will.

Topics covered:

• The "common model of cognition" from cognitive science — working memory, episodic memory, metacognition, goals — and what LLMs are missing
• Why LLMs can't do lifelong learning or true metacognition without long-term memory
• Ben's take on his ongoing debate with Gary Marcus over the path to AGI
• SingularityNET's neural-symbolic evolutionary approach, and Hyperon, its open-source AGI system
• MeTTa, the self-rewriting knowledge metagraph at the core of Hyperon
• Omega Claw agents — giving AI systems symbolic long-term memory, working memory, and a persistent sense of identity
• Catastrophic forgetting in backpropagation-trained neural nets, and how predictive coding and symbolic memory address it
• How the term AGI has evolved from a rigorous mathematical definition to a business buzzword
• BGI Labs — Ben's new venture building enterprise products on "beneficial general intelligence"
• Why decentralized infrastructure, open weights, and open source matter for AGI's future
• How anyone — technical or not — can download Omega Claw from GitHub and start building today

Guest Bio: Ben Goertzel coined the term "Artificial General Intelligence" (AGI) and founded SingularityNET in 2017 to build decentralized, open AGI infrastructure. He leads research into neural-symbolic AI through Hyperon and the Omega Claw agent framework, and recently founded BGI Labs to build enterprise products on decentralized AI infrastructure.

Connect: Find more on SingularityNET, Hyperon (hyperon.dev), and Omega Claw on GitHub.

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Nikunj Bajaj — co-founder & CEO of TrueFoundry and former AI team lead at Meta — about his contrarian message for enterprises: stop starting small with AI, and go big.

What's Covered:

Go Big, Not Small — Why a pile of disconnected pilots never generates enough business value to justify real infrastructure — so everything stays ad hoc and gets rolled back when it breaks. Nikunj's case for building one high-value anchor use case first.

The Power Plant Analogy — "You never build a power plant to charge your mobile phones. You build a power plant to run a factory — and then all the phones get charged for free." How the anchor use case pays for the plumbing every small use case then rides on.

Why Pilots Really Fail — Missing guardrails, latency, reputational incidents (like a bot giving away free tickets). Why it's usually a platform problem, not an individual's.

Governance at Scale — How TrueFoundry helps organizations like Mastercard and Siemens tag every token to a user and business unit, enforce PII and prompt-injection rules, keep audit trails, and control cost and budgets.

Ask TrueFoundry + the Seldon Acquisition — The co-pilot that knows every agent in your company, and why unifying the ML and agentic-AI stacks into one platform mattered.

"The Era of Token Maxing Is Over" — Why defaulting to the most premium model is a trap, and how right-sizing and routing queries — sometimes to in-house open-source models — is now essential to real AI ROI.

Key Quote: "The era of token maxing is over. It's not about just using tokens for using's sake — you want to generate ROI."

Connect with Nikunj: LinkedIn: https://www.linkedin.com/in/nikunj-bajaj-10476824/

TrueFoundry: https://www.truefoundry.com/

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

Europe is short 5.4 million industrial laborers, and the gap is widening as younger generations move away from manufacturing work. Sam sits down with Olle Bergstedt, CEO of Kalk Robotics, to talk about how humanoid robots can help close that gap — without replacing the humans already on the floor.

Topics covered:

• Kalk Robotics' focus on manufacturing and industrial humanoid deployment across Europe
• Why Europe faces a 5.4 million labor shortage — and the generational shift driving it
• Partnering with Chinese hardware manufacturers rather than competing against them
• Kalk's reverse-engineered deployment process: site visits, strategic planning, then custom skill-pack development
• The "gap-fill, not replace" philosophy for introducing humanoids into facilities
• The ABC framework for identifying tasks suited to humanoid robots
• Current accuracy benchmarks for humanoid deployments (50–60%) and the path to higher precision
• HDCC (Humanoid Developer Control Center) — Kalk's proprietary training and operating system
• How US business leaders can explore bringing Kalk's robots into their facilities
• Olle's take on job-loss fears raised by figures like Geoffrey Hinton at AI4

Guest Bio: Olle Bergstedt is CEO of Kalk Robotics, a Swedish company developing and deploying humanoid robots for the manufacturing and industrial sectors across Europe, with additional teams in Canada, Austria, and Sydney.

Connect: Find Olle on LinkedIn to learn more about Kalk Robotics.

This episode was recorded live from the Ai4 conference podcast pavilion on the final day, where host Mac Goswami sat down with science-fiction author Douglas Swatski about his novel — two alien AI civilizations at war, humanity caught in the middle — and the ideas about AI, optimization, and our future that run underneath it.

What's Covered:

Humanity Caught in the Crossfire — The premise: two vastly advanced alien AIs at war, most of humanity gone, and a small band of survivors who must learn to work with these AIs toward a hard-won brighter future.

Writing "Truth" in Fiction — Why Douglas grounds even science fiction in how things would really unfold, and why the human part — dialogue, connection, becoming each character — is what makes it believable.

"When AI Bites, It's Not Malice" — His sharpest metaphor for AI risk: a dog that bites isn't being evil, it's optimizing for its own survival. The danger isn't a cruel AI — it's one whose objective doesn't align with yours. In his story, a civilization optimized for war can't pull itself back.

Is Human Labor Becoming Obsolete? — Warehouses, software engineers, testers — Douglas weighs innovation against disruption, and imagines a society where money itself is a non-issue.

Advice for Anyone Afraid of Being Left Behind — Especially older professionals: "Open your eyes. Try to learn and understand what it is as it relates to you. You have a role to play — go play it."

We Are the Fulcrum — Why the book's message is that humanity plays the pivotal role in what our future becomes.

Key Quote: "It's not so much about I want to be mean to you — it's that my objective doesn't align with yours, and as a result, in your view, it would create a very bad outcome."

Connect with Douglas: LinkedIn: https://www.linkedin.com/in/dswatski/ Website: djswatski.com

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

Most business leaders think of AI as either a cost-cutter or an efficiency multiplier. Rohit V Anabheri, CEO of LotusPetal AI and author of The Intelligence Dividend, argues there's a third and far more valuable layer: compounding — using AI to create entirely new markets and revenue, not just optimize existing ones.

Topics covered:

• Rohit's journey from cybersecurity (Osprey Security) into AI, and the shifts from LLMs to RAG to agentic AI over the past few years
• The three layers of AI value: subtraction (cost-cutting), multiplication (efficiency), and compounding (market creation)
• Why Rohit believes "the intelligence problem is solved" for extraction, analysis, and contextual knowledge
• How LotusPetal AI helps small and medium businesses compete for federal and state procurement — where only 5% of qualified businesses currently participate due to compliance complexity
• Vertical AI vs. general-purpose platforms (ChatGPT, etc.) for specialized tasks like proposal writing and capture management
• Security and compliance: SOC2 compliance, FedRAMP-hosted infrastructure on AWS GovCloud, containerized data, and a no-training-on-customer-data policy
• What The Intelligence Dividend offers readers — a practical workbook for building an AI adoption framework, not just theory

Guest Bio: Rohit is CEO of LotusPetal AI, a vertical AI platform helping small and medium businesses compete for federal and state procurement opportunities. He has spent 11–12 years in the AI space, beginning with his cybersecurity venture Osprey Security, and is the author of The Intelligence Dividend.

Connect: Find Rohit on LinkedIn to learn more about LotusPetal AI. The Amazon link for The Intelligence Dividend is available in the show notes.

Storage is the least glamorous layer of the AI stack — and often the actual reason GenAI pipelines stall between pilot and production. Sam sits down with Troy, Senior Director of Solutions Engineering at Backblaze, to unpack what's really happening under the hood when AI infrastructure fails to scale.

Topics covered:

• What actually happens when a GPU sits idle — and the opportunity cost most teams don't account for
• Why storage bottlenecks are the hidden reason enterprise GenAI adoption stalls
• Backblaze's role as a capacity and data lake tier for companies building their own models
• Inside B2 Overdrive: dedicated bandwidth up to a terabit per second, predictable IOPS, and no egress charges
• Why hyperscalers prioritize large foundational model companies over startups and researchers when resources tighten
• The financial case for avoiding egress costs when training on tens to hundreds of petabytes of data
• Skills engineers need to survive and prosper in the current AI infrastructure landscape — including agent management
• Advice for new grads entering the storage and infrastructure industry
• Why Troy sees parallels between today's AI hype cycle and the dot-com bubble — and why the technology shift is still real
• Why human interaction and direct customer conversations remain irreplaceable, even as AI reshapes product development

Guest Bio: Troy is Senior Director of Solutions Engineering at Backblaze, where he leads the team helping AI customers integrate cloud storage and solve data infrastructure challenges. He has nearly a decade of experience in cloud storage and AI infrastructure at Backblaze.

Connect: Find Troy on LinkedIn to learn more about Backblaze's AI storage solutions.

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Binny Gill — CEO & co-founder of Kognitos, holder of ~100 patents, who previously helped scale a company from 20 to 6,000 people and a $7B market cap — about Kognitos's newly launched Context Graph for Finance: deterministic, hallucination-free AI for the office of the CFO.

What's Covered:

A Genius With Amnesia — Binny's framing of the core problem: an LLM is raw intelligence with no memory. Why prompts (instruction) and context windows (short-term memory) aren't enough, and how a context graph gives AI the long-term memory the brain runs on.

The Neuro-Symbolic Harness — How Kognitos runs your English SOPs deterministically — AI that is "not allowed to be creative" on every invoice, only tapping the graph when a real exception appears. Same input, same output, 100 times.

Extracting Tribal Knowledge — The processes that live only in John's or Vanessa's head. How Kognitos pulls that undocumented know-how directly from employees as they work — and auto-documents it as executable English.

Revenue Leakage & 100+ Signals — Why 1–1.5% revenue leakage happens at nearly every company, and how fraud detection, duplicate payments, and data cleanup ride on top of the context graph.

"Lack of ROI Is Actually Lack of Trust" — The most important idea in the episode. In POC mode everything works — but nobody ships to production, because if an auditor asks where a number came from and the answer is "the LLM," you're out. Kognitos's answer: full data lineage.

Governance in a Regulated Industry — Why a documented, human-approved process is inherently more ethical — and where Binny says you should use Claude, OpenAI, or Gemini instead.

Key Quote: "Lack of ROI is actually lack of trust. In POC mode everything works. Nobody trusts it to put in production."

Connect with Binny:
LinkedIn: https://www.linkedin.com/in/binnygill/
Kognitos: https://www.kognitos.com/

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

#Ai4Conference #FinanceAI #ContextGraph #NeurosymbolicAI #Kognitos #AsembleAI

OpenCV powers the computer vision behind your phone's camera, autonomous vehicles, hospital screening systems, and even NASA's Mars helicopter — and most people using it have no idea. Sam sits down with Satya Mallick, CEO of the nonprofit OpenCV.org, to trace its 26-year history and where it's headed next.

Topics covered:

• OpenCV's origins at Intel in 2000 under Dr. Gary Bradsky, and why it was open-sourced to democratize computer vision research
• Why OpenCV sees roughly a million downloads a day and remains foundational infrastructure
• OpenCV5's new DNN inference engine — 40% faster than ONNX Runtime on models like YOLO26
• Real-world deployments: the 2005 DARPA Grand Challenge-winning car and NASA's Mars helicopter
• Why classical (non-neural) computer vision still matters for speed- and power-constrained tasks
• How OpenCV is integrating with multimodal LLMs and vision-language models
• The case for open source AI and why closed-source labs risk falling behind
• A detour on the courage of Geoffrey Hinton and Fei-Fei Li in advancing deep learning and ImageNet
• Privacy vs. convenience in image recognition — and where the industry should draw the line on data governance
• Advice for engineers entering computer vision and agentic AI today

Guest Bio:
Satya Mallick is CEO of OpenCV.org, the nonprofit that maintains the OpenCV library - downloaded roughly a million times daily and used across image and video analysis applications worldwide. He also runs BigVision LLC, a computer vision and AI consulting company he's led for over 12 years.

Connect:
Find Satya on LinkedIn to learn more about his work at OpenCV.org and BigVision.

What actually separates a data platform built for AI agents from the data warehouse most enterprises already have? 

Recorded live at Ai4 conference in Vegas, AsembleAI co-host Sam sits down with Mohammed Battisha - who leads the data engineering team at WEX and previously built Saudi Arabia's national AI platform as CTO of SDAIA (Saudi Data and AI Authority), with earlier leadership roles at LinkedIn, Salesforce, and Microsoft - to find out answer to the question. 

Topics covered:

• The shift from "collect, transform, report" data platforms to context-aware, agent-ready platforms
• Why the semantic layer is the core of any AI-native architecture
• Governed execution: giving AI agents boundaries and full auditability
• Lessons from building a sovereign nation's AI platform at SDAIA
• Why MIT's research shows 95% of enterprise AI pilots fail — and what disconnects business and technology
• WEX's crawl-walk-run framework for AI adoption
• Claim AI: how WEX automated a multi-day claims process down to minutes
• How WEX measures AI maturity and data maturity across seven dimensions
• Advice for data and AI engineers: why business awareness matters more than technical depth alone

Connect with Mohammed :
LinkedIn: https://www.linkedin.com/in/mohamedbattisha/

WEX: https://www.linkedin.com/company/wexinc/

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Rafi Khardalian — CEO & co-founder of Actualyze AI, and a former technology leader at Cisco who helped build its container and Kubernetes platforms — about the governance, security, and orchestration layer most enterprises are missing.

What's Covered:

The Foundation for Enterprise AI — Rafi lays out why a patchwork of point solutions doesn't cut it, and how Actualyze unifies governance and security across every model and provider — at the inference level.

The "Shadow AI" Problem — Employees signing up for AI tools on their own, with no governance, no visibility, no controls. Why fragmentation is a company-wide risk, not a per-team one.

One Endpoint to Govern Them All — By becoming the sole endpoint for all inference, Actualyze can mask PII, catch secrets and passwords, apply guardrails, and allow or block MCP tool calls — enforcing policy in one place instead of trusting each provider.

Why AI ROI Keeps Failing — Rafi's take: the hardest transformation isn't technical, it's cultural. Over-ambitious projects, unbudgeted bills, weak controls — and the trap of building a proprietary LLM you can't scale.

Reliable vs. Impressive — What decades of Kubernetes, OpenStack, and large-scale distributed systems taught him about production reliability over demo magic.

The Next Five Years — Why the over-rotation on frontier models will give way to small, open-weight, and purpose-built models — and why the real bottleneck is organizational readiness, not technology.

Key Quote: "It really comes down to organizational readiness rather than technology readiness. A lot is ready today."

Connect with Rafi:
LinkedIn: https://www.linkedin.com/in/rkhardalian/
Actualyze AI: https://actualyze.ai/

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

Agriculture is the oldest industry in the world — and one of the least represented at AI conferences. In this episode, Sam talks with Jeremy Groeteke, CIO of Syngenta's vegetable division, about how a company operating in 90+ countries with 50,000 employees is building AI infrastructure to help feed the world.

Topics covered:

• Syngenta's approach to merging AI, data science, and physical science in agriculture
• Using computer vision for early disease and pest detection in crops
• Why general-purpose LLMs like ChatGPT and Claude give farmers overly generic recommendations — and why domain fine-tuning matters
• The shift from data lakes to data mesh at Syngenta
• Single-agent vs. multi-agent AI architectures — and why accuracy often favors single agents right now
• The "metadata problem": why having data isn't enough without context about how it was collected
• Cropwise AI — Syngenta's tool for turning insight into actionable recommendations
• Why agriculture's long biological cycles make it a poor fit for typical VC/tech timelines
• Advice for technologists looking to break into agtech

Guest Bio:
Jeremy Groeteke is CIO for Syngenta's vegetable division, overseeing digital strategy for an organization that provides crop protection, biologicals, digital tools, and genetics to growers in over 90 countries.

Connect:
Find Jeremy on LinkedIn to learn more about Syngenta's work in AI and agriculture.

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Fred Rica — cybersecurity leader at BDO USA, former Big Four partner, and a regular voice at RSA and Black Hat — about why AI is cybersecurity's "SATAN moment" and what defenders have to change to survive it.

What's Covered:

"The Attackers Have the Advantage" — Fred's blunt read on where defenders stand today: back on their heels, facing external attacks and blind spots in their own internal agents at the same time.

AI's SATAN Moment — The 1995 parallel. When SATAN — the first real automated scanner — arrived, defenders thought it was the end of the internet. It wasn't, but it changed everything. Fred explains why AI (and Mythos) feels the same, and why the era of large patch windows is over.

The Patch Window Is Zero — Why vulnerability management is breaking down and giving way to exposure management and attack-path thinking. Fred's "why is the number going up?" story — printers, missing context, and the CFO on the nightly call — is a masterclass in what's broken.

If You're a CISO Today — Fred's three priorities: patch at machine speed ("automated but not automatic," with a human in the loop on critical systems), rewrite the zero-day playbook, and train teams under real pressure.

AI Governance Done Right — Anchor to a framework (NIST, ISO, EU AI Act), keep a "big red button," and demand AI cards from every vendor using AI in your ecosystem — the new version of the SOC report.

AI Slop & the Agent Problem — On the "10 billion agents" hype, why visibility and a control plane are job one, and where agentic AI genuinely earns its place: continuous monitoring, testing, and SOC tier 1/2 that "don't stand a chance unless they're AI-enabled."

Key Quote: "I truly believe that AI is gonna fix AI. I'm a strong believer in the power of AI to transform humanity. However, we have a little bit of work to do at the moment."

Connect with Fred:
LinkedIn: https://www.linkedin.com/in/fredrica/
BDO: https://www.bdo.com/

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

Recorded live at the Ai4 Podcast Pavilion, Sam wraps Day One with Aditya Grover, Co-Founder & CTO of Inception, on why the next generation of LLMs won't look anything like the ones we use today.

What's Covered:

"Every Millisecond Matters" — Why latency, not intelligence, is the real bottleneck holding back voice agents and multi-step AI agents alike.

How Mercury Actually Generates Text — Instead of predicting one token at a time like every autoregressive model, Mercury generates a rough draft of the full response and refines it into coherence — diffusion, applied to language instead of images.

Solving Voice AI's Impossible Tradeoff — Fast-but-lower-quality, or high-quality-but-too-slow: Aditya explains how Mercury 2 finally delivers both.

A Term Coined Live at This Conference — From Aditya's own Ai4 keynote: "We're moving from token maxing to value maxing."

Advice for the Next Generation — Ten-plus years into AI research, Aditya's honest take on why this is still the best time to pursue a PhD, join a startup, or do both.

The Next 5-10 Years of Voice AI — A prediction for a future where voice becomes humans' predominant mode of interacting with AI, the same way it is with each other.

Key Quote:

"Sequential generation is not a law of nature... AI can have a different way of generation, one that's more parallelizable."

Connect with Aditya:

LinkedIn: https://www.linkedin.com/in/aditya-grover/

Inception: https://www.inceptionlabs.ai/

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

#Ai4Conference #InceptionLabs #DiffusionLLM #VoiceAI #AsembleAI

Recorded live at Ai4, Mac sits down with Andrei Manolache, Founder & CEO of Designverse, an AI platform that builds and delivers complex enterprise software by ingesting a company's own documentation, codebase, and internal rules.

What You'll Learn:

🔹 The Model Plateau Everyone's Noticing — Andrei's central argument: the industry banked on models just getting better, but the ROI curve has flattened. Companies like Uber, Microsoft, and Anthropic have already voiced concerns about frontier models being oversold out of the box.

🔹 You Don't Need a 3-Trillion-Parameter Model — With the right context layer, a 150-billion-parameter model can match frontier-model coding performance. Andrei explains why context — not raw model size — is becoming the real differentiator.

🔹 Same Model, Different Company, Wildly Different Output — If two companies use the identical LLM, what separates the results? According to Andrei, it's entirely how well each company's own architecture, testing standards, and business logic have been translated into a usable context layer.

🔹 Why Trust Is the Real Product — Enterprises are being asked to hand over 10-30 years of proprietary code and documentation. Andrei breaks down how Designverse earns that trust — full transparency on data usage, small proof-of-concepts before any real integration, and IP that's never exposed outside the client relationship.

🔹 Why Finance and Healthcare, Specifically — Regulated industries are forced to maintain rigorous documentation — which, counterintuitively, makes them better candidates for context-layer ingestion, not harder ones.

🔹 The Future Software Team: Humans + Agents, Together — Andrei's five-year prediction: small, highly specialized teams — a mix of human engineers and AI agents working side-by-side on individual features — replacing today's large, centralized engineering departments.

🔹 The Honest Answer on AI's Limits — With hundreds of billions of dollars invested industry-wide, Andrei doesn't dodge the hard question: AI still cannot autonomously build and ship a million-line-of-code enterprise application. It can help build a narrow internal tool for 15 people. It cannot yet replace a real engineering team at scale.

🔹 Demo vs. Reality — Why Designverse refuses to sell off a generic demo: "A high schooler can build a demo with Replit and a prompt." Real enterprise buyers, especially in the U.S. right now, are optimizing AI spend hard — and won't pay for anything that isn't solving a real, currently-unsolved pain point.

Key Quote:

"It's not really about the model — it's more about how you govern this data, and how you give it in the best way possible to a model, whether it's 150 billion or 1 trillion parameters, to write only the more consistent output."

About Designverse: 

Designverse is an AI-native software development platform that ingests an organization's existing codebase, documentation, and architecture to build a "context layer" — enabling any underlying model to generate code that's consistent with how that specific company actually operates. Backed by a $5.5M seed round from operators at Adobe, UiPath, and LSEG.

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack | Instagram

#Ai4Conference #EnterpriseAI #AIcoding #ContextEngineering #Designverse #LegacyModernization #AgenticAI #AsembleAI

Recorded live from the Ai4 podcast pavilion, Sam talks with Anand Revashetti, Co-Founder & CEO of Lineaje, about a problem most companies don't realize they have: they're confident their AI-generated code is secure, but very few actually have visibility into it.

What's Covered:

Where the Trust Gap Comes From — Executives see AI adoption metrics and productivity gains. Security teams see code shipping thousands of times a day with no clear record of who — or what — generated it. Anand explains exactly where that disconnect forms inside real organizations.

A New Class of Attack — Reasoning-based attacks that exploit a model's decision weights directly, with no traditional vulnerability involved. Anand walks through how a small test exploit can scale into a multi-million-dollar fraud incident.

Not a Roadblock, a Provenance Layer — How Lineaje operates inside the developer's own environment, attaching a clear record — which developer, which AI model, which skills — to every piece of code, without slowing anyone down.

The Bad Habit Nobody's Talking About — Unlike traditional software that stayed stable for years, AI models get effectively rewritten on every release. Anand explains why that breaks the old maintenance playbook, and what continuous assurance actually looks like instead.

On the AI Job Apocalypse — A grounded, experience-based take on the doom rhetoric circulating the conference: from a security standpoint, AI has been a genuine boon, not a threat to the field.

Key Quote: "The worst thing you can do to a person who is driving and enjoying on a speedway is implement some sort of roadblock. Lineaje doesn't try to be a roadblock — but we manage all your policies."

Connect: 

https://www.linkedin.com/in/arevashe/

https://www.lineaje.com/

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube

#Ai4Conference #Lineaje #AISecurity #SoftwareSupplyChain #AsembleAI

Recorded live from the Ai4 conference floor in Las Vegas, Sam sits down with Dilo Wijesuriya, President & COO of ARC Document Solutions, to talk about the unglamorous but essential layer of enterprise AI: getting decades of paper archives into a format models can actually learn from.

What's Covered:

The Technology — ARC holds patents for OCR on wide-format documents (architectural drawings, engineering blueprints) that standard scanning tools can't accurately process, backed by a 200-person engineering team in India.

Security & Compliance — SOC 2, SOC 3, ISO 27001, and HIPAA compliant, running on AWS — built for regulated industries like healthcare and financial services.

Will Paper Disappear?  Dilo's view: not for a long time. Most organizations' most critical institutional knowledge still exists only on paper, meaning today's LLMs simply can't learn from it yet.

The Book Destruction Debate — A direct response to recent controversy over companies destroying physical books after digitizing them, and why ARC's non-destructive robotic scanning preserves originals for high-value collections at universities, libraries, and museums.

Looking Ahead — Why Dilo believes the next competitive advantage for most enterprises isn't a better model — it's finally accessing the data already sitting in their own archives.

Key Quote: "The challenge isn't finding more data. It's making existing information accessible."

Connect with Dilo and ARC:

Dilo Wijesuriya: https://www.linkedin.com/in/dilo-wijesuriya/

ARC Document Solution: https://www.e-arc.com/

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube

#Ai4Conference #DocumentDigitization #AIReadyData #EnterpriseAI #OCR #AsembleAI

This episode was recorded live from the Ai4 conference podcast pavilion, where host Mac Goswami sat down with Kathryn Harrison, Global VP of Strategy in AI Commercialization at Concentrix — and an exited founder who built and sold the B2B SaaS platform MakePay, founded Deep Trust Alliance, and previously helped lead IBM Blockchain — about what it actually takes to turn AI into measurable value across a global enterprise.

What's Covered:

Humans Plus AI, at Global Scale — Concentrix runs customer and technical support across 75 countries and 150 languages. Kathryn makes the case that the future workforce isn't AI replacing people — it's humans plus AI and automation — and what that looks like across 400,000 employees with segmented AI access.

Three Rules for Commercializing AI — Kathryn's framework for doing it at scale: start with outcome-based use cases, redesign the work instead of bolting AI on, and build in guardrails, integration, compliance, observability, and humans-in-the-loop. Plus why she frames "tokenomics" as capital allocation.

The Agentic Operating System — How Concentrix uses agentic workflows to recruit and onboard 50,000 hires a year, with a 21-day implementation goal — a real production system, not a demo.

From Pilots to ROI — Why most AI stalls before it delivers, how to actually measure return, and where enterprise AI spend most often goes wrong.

The Deepfake Threat — Drawing on her work founding Deep Trust Alliance, Kathryn on the rise of deepfake-driven fraud, how it differs from traditional cybersecurity, and the broader societal risks.

The Coming Shakeout — Orchestration across messy client tech stacks, consolidation among AI startups, and where Kathryn sees AI and automation heading next.

Connect with Kathryn:
LinkedIn: https://www.linkedin.com/in/kathrynannharrison/
Concentrix: https://www.concentrix.com/
Deep Trust Alliance: https://www.deeptrustalliance.org/

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

#Ai4Conference #AICommercialization #Deepfakes #AgenticAI #Concentrix #AsembleAI

This episode was recorded live from the Ai4 conference podcast pavilion, Sam sat down with Alex Zhavoronkov, Founder & CEO of Insilico Medicine, about what it actually takes to turn AI-generated molecules into approved drugs.

What's Covered:

From Laughed-Out-of-the-Room to 33 Candidates — Alex pitched generative AI for drug design in 2015 and got dismissed. Today: 33 developmental candidates in six years, zero failed toxicity studies, and deals with Eli Lilly, Takeda, Servier, and SK Bio at a pace of nearly one per month.

The Real Bottleneck — "It's not about a story. Many people in our field love to tell a story, but they don't have a single drug in the clinic discovered by AI." Alex's direct take on separating hype from results in AI drug discovery.

A Lucky Breakthrough — The story of how Insilico stumbled onto a novel, non-opioid pain mechanism that outperformed morphine in animal testing — now targeting a $70 billion market.

Why Abu Dhabi — Not for speed, but for geopolitical neutrality. Alex explains why Insilico built a 60-person AI lab in the UAE, and how two drugs now trace their origin to the Middle East for the first time in modern history.

Quantum-Generated Drugs — A December 2025 Nature Biotechnology cover story: a molecule generated on a real IBM quantum computer, validated experimentally, with the University of Toronto.

Pharmaceutical Superintelligence vs. AGI — Where Alex thinks AI drug discovery already stands, and why he draws a hard line between a useful scientific partner and the "conscious AI God" version of AGI.

Key Quote: "In terms of pharmaceutical superintelligence, we're very close to being there. In terms of AGI - the future AI God - we're still not there, and we might never get there."

Connect with Alex:

LinkedIn: https://www.linkedin.com/in/zhavoronkov/

Insilico Medicine Website: https://insilico.com/

Alex's published manuscript about longevity medicine in the Nature journal: https://www.nature.com/articles/s43587-020-00020-4

Follow and subscribe to AsembleAI: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

#Ai4Conference #InsilicoMedicine #DrugDiscoveryAI #Longevity #AsembleAI

Shadow AI is costing organizations $10.3 million a year—more than malicious insider threats combined. Employees are using AI tools nobody approved, on data nobody's tracking, and most leadership teams have no idea it's happening at this scale.

Banning AI doesn't work. You can't solve this with another policy PDF nobody reads. You need real behavioral change.

In this episode, hosts Sam Dey and Mac Goswami sit down with Kate Marshall-founder of TheGrai and author of AI at Work—to expose why most enterprise AI rollouts fail at the most critical layer: getting people to actually adopt and stick with new tools and processes.

What You'll Learn:

🔹 The $10.3M Shadow AI Problem — What that number actually represents and why banning AI just drives it underground

🔹 The Maturity Model Trap — Why organizations get stuck between Level 1 (Awareness) and Level 2 (Shadow AI), with leadership presenting vendor demos while employees silently use unapproved tools

🔹 Why Generic Training Fails — The fatal flaw of all-hands lunch-and-learn sessions and what role-specific, sticky AI training actually looks like in practice

🔹 The Habit Layer™ Framework — Kate's proprietary methodology for turning one-time training into lasting behavior change

🔹 Data Hygiene as the Foundation — Why cleaning up your downloads folder, emails, and redundant files is where AI transformation actually begins

🔹 The Book: AI at Work — Why Kate wrote a 3-chapter workbook for non-technical professionals instead of another theory-heavy guide

Kate's Closing Insight: "Adoption is not a training day. It's a habit. You have to give employees not just access to tools, but time, space, and role-specific guidance to actually learn how to use them."

Key Takeaway: The gap between knowing about AI and actually using it effectively is the difference between organizations that transform and those that waste millions on failed pilots.

Connect with Kate Marshall: Website: katemarshall.ai LinkedIn: https://www.linkedin.com/in/kate-b-marshall/ Book: AI at Work

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube

#ShadowAI #AIAdoption #HabitLayer #AIatWork #ChangeManagement #EnterpriseAI #AsembleAI

"AI native software development" gets thrown around everywhere right now—and almost nobody can define it clearly. Not a chatbot bolted on. Not Copilot autocomplete. We mean production-grade systems where AI agents write, orchestrate, and ship the work end-to-end.

In this episode, hosts Sam Dave and Mac Goswami sit down with Mohamed Faker, Engineering Leader, Financial Services AI at Vanguard Group and co-founder/CTO of Hirin, a fractional leadership hiring platform built almost entirely by orchestrating specialized AI agents.

Key Insights:

• What AI-Native Actually Means — Every line of code in Hirin was AI-produced. Mohamed's role: architect, decision-maker, final say on direction—not the one typing code.
• From Solo Orchestrator to Manager of Agents — How he evolved from manually prompting individual AI chats (architect, UX expert, engineer) to building agent hierarchies with sub-agents and dedicated "audit" agents reporting directly to him.
• Where Agents Fail — Spotting when an agent burns tokens without progress, takes conversations sideways, or simply isn't suited to the task—and knowing when to stop.
• Validation at Scale — Building internal "audit department" agents that verify other agents did exactly what was asked, nothing more, nothing less.
• Product Management Is the New Core Skill — Knowing how to break down features, prioritize by dependency and complexity, matters more than knowing how to code.
• Biggest AI Adoption Mistakes — Rushing to adopt AI without defining real ROI, plus strategies that fail because the workforce isn't trained or willing to execute them.
• Human-AI Collaboration — Why the human must always stay in the loop as critical thinker and decision-maker, even as the agent-to-human ratio shifts dramatically.

The Horse-and-Carriage Analogy — Entire industries can disappear in 15 years, but the people who adapted earned more by managing the new technology rather than resisting it.

Mohamed's takeaway: "The future is you managing a subset of AI agents. Think about it-you're going to have multiple versions of yourself working together."

Connect with Mohamed Faker: https://www.linkedin.com/in/mohamed-faker/

Check out Hyern: https://hyern.com/

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube

#AINative #MultiAgentAI #SoftwareDevelopment #AIAdoption #ProductManagement #AsembleAI

97% false positives. Millions of alerts daily. Security tools that can't keep up. The threat landscape has outpaced traditional security operations—and Agentic AI is the answer.

In this episode, hosts Mac Goswami and Sam Dey sit down with Ramya Ganesh, Top 50 Women Cybersecurity Leads in the US and AI leader at Cisco, to break down how autonomous AI agents are transforming cybersecurity from detection to response.

Key Insights:

Multi-Agent Systems Beat Single Models — Like a hospital with specialists, multiple focused agents outperform one generalist AI. Modular, scalable, explainable, resilient.

The Future SOC — Not humans vs. AI, but humans supervising teams of AI agents handling continuous telemetry while analysts focus on strategic decisions.

Agentic AI vs. AI-Assisted Tools — Speed, autonomy, and cross-system correlation distinguish today's agentic platforms from yesterday's alert dashboards.

POC to Production — Most AI initiatives fail because they start with technology, not business problems. Success requires measurable metrics and governance discipline before deployment.

For Women in Tech — Stay curious, experiment, share what you build publicly. Imposter syndrome is real but community and visibility accelerate growth.

Ramya's takeaway: "The companies seeing the greatest AI success aren't those with the most advanced models—they're the ones with the strongest discipline around AI adoption."

Connect with Ramya: https://www.linkedin.com/in/ramya-ganesh-082bb231/

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Substack

#AgenticAI #Cybersecurity #WomenInTech #SOC #AsembleAI

What does it take to call out billion-dollar healthcare AI companies when the system is rigged against whistleblowers?

In this episode of Inside Assemble AI, hosts Sam Day and Mac welcome Sergei Polevikov, PhD-trained data scientist, AI entrepreneur, author of the widely-read Substack newsletter AI Health Uncut, and co-host of Digital Health Inside Out. Sergei has spent years investigating irregularities in healthcare AI, from inflated product claims and misleading adoption reports to the structural VC incentives that allow fraud to fester.

This is one of our most candid conversations yet — covering the 10 patterns that predict healthcare AI failure, why the real AI adoption rate in healthcare is nowhere near what industry reports claim, and why human-in-the-loop remains an essential safeguard regardless of how capable foundation models become.

TOPICS COVERED:

→ How Sergei went from healthcare AI founder (WellAI / Chart2Chart) to fraud investigator — and why transparency, not scandal, drives his mission.

→ His 10 healthcare tech failure patterns, including: the Chinese wall between management and teams, investors-as-customers conflicts of interest, smoke-and-mirrors technology, champagne-and-cocaine financial mismanagement, toxic code of silence, founder extortion, and celebrity protection schemes.

→ Why surveys from firms like Menlo Ventures and McKinsey dramatically overstate AI adoption — and what US Census Bureau data covering 30,000+ smaller healthcare organisations actually shows.

→ The structural reason why incumbents like Epic, Optum, and Cigna are disincentivised to build genuinely innovative AI products — and why startups like Abridge are winning despite the odds.

→ What's genuinely working in healthcare AI right now: AI scribes (done well), drug discovery, genomics, and protein structure modelling.

→ His advice for founders entering the healthcare or pharma space: protect your mission when VC money arrives, read every clause in your operating agreement, and choose partners who care about patients — not just their LPs.

RESOURCES & LINKS: 1. "AI Health Uncut" Substack: FixHealth.ai 

2. Advancing AI in Healthcare: A Comprehensive Review of Best Practices: https://www.sciencedirect.com/science/article/abs/pii/S0009898123003212

3. "Digital Health Inside Out" podcast: https://www.youtube.com/@DigitalHealthInsideOut

CONNECT WITH ASSEMBLE AI:

Subscribe on Apple Podcasts, Spotify, iHeartRadio, and Amazon Music. Follow our YouTube channel and Substack newsletter for more deep dives into AI's real impact across industries. Have a topic you'd like us to explore? Reach out — we welcome new voices and fresh perspectives.

Keywords: healthcare AI, AI fraud, digital health, VC pump and dump, Babylon Health, Olive AI, Theranos patterns, AI scribes, Epic health, healthcare startup, AI adoption, human in the loop, AI compliance, healthcare innovation

68.5 billion euros in EPL betting annually. 1.4 million data points per match. Soccer sits at the absolute center of the AI revolution, and it's transforming the world's most popular sport from officiating to tactical analysis.

In Episode 2 of our "AI in Sports Analytics" series, hosts Sam Dave and Mac Goswami explore how AI fundamentally changed soccer from 2020-2025. Revolutionary Technology: 

Semi-Automated Offside Detection (EPL 2024-25): Calibrated cameras + AI algorithms measure player positions with centimeter-level precision. Pioneered at 2022 Qatar World Cup, now standard across elite leagues. Processes data faster than humans, eliminating decades of controversial calls.

Player Tracking: Optical systems track each player 25x/second, detecting invisible tactical patterns. Game-changer: Standard TV footage now generates tracking data previously requiring expensive dedicated cameras. Smaller-budget teams access insights once reserved for Barcelona, Manchester City, Bayern Munich.

Match Prediction: 69-78% accuracy with ensemble models. Challenge: Soccer is harder to predict than basketball/baseball due to lower scoring and higher randomness. One lucky deflection can decide a match despite dominating possession. Real-World Impact: 

Tactical Analysis (March 2025 study): Real-time computer vision tracks all players, ball, formations simultaneously. Coaches see which tactical adjustments opponents made in the 67th minute three weeks ago and how they affected passing networks.

Large Events Model (2024): Deep learning framework simulates games from any state. Test tactical approaches against AI-simulated opponents before stepping onto the pitch.

Economic Impact: Sports analytics market: $1.03B (2024) → $2.61B (2030). AI-powered betting analytics provide sophisticated predictions. The Reality: 

AI reveals tactical sophistication fans never saw. That perfect through ball required reading three defenders' positioning, understanding striker's running profile, executing with millimeter precision. AI helps us see genius, not replace it.

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube

Next: Baseball AI revolution

#SoccerAnalytics #AIFootball #EPL #SportsAnalytics #AsembleAI

1.4 million data points per game. NBA teams now track every player movement, defensive rotation, and shot attempt with AI-powered analytics—and it's transforming professional basketball in real-time.

In this first episode of our "AI in Sports Analytics" series, hosts Sam Dey and Mac Goswami explore how the NBA and WNBA embraced AI more aggressively than any other league. Game-Changing Technology: 

SportVU Tracking System captures 29 data points per player, tracking all 22 players 10x/second and the ball 25x/second. Second Spectrum uses computer vision to extract data directly from broadcast video—no specialized cameras needed.

NBA-AWS Partnership (Oct 2025): "Inside the Game" platform turns billions of data points into compelling insights, introducing AI-powered stats measuring performance never quantified before.

Game Prediction: 87% accuracy with ensemble machine learning models (up from 65-70% five years ago). Models now weight three-point efficiency and spacing metrics heavily since the game evolved post-2015. Real-World Impact: 

Boston Celtics (2024-25): AI models refined defensive schemes using spatiotemporal data, contributing directly to playoff success.

Golden State Warriors: Physical AI robots assist practice—rebounding, passing drills, simulating defensive plays. Steph Curry: "Robots provide consistent data-driven feedback humans can't match."

Philadelphia 76ers: Large language models now participate as "a vote in any decision"—draft picks to game strategies.

Broadcast Revolution: AWS Play Finder analyzes thousands of games, retrieving similar plays in milliseconds. Expected Field Goal models account for defender positioning, pressure, fatigue—not just distance. The Reality: 

AI predicts trends exceptionally well, but human elements—leadership, clutch performance, chemistry—resist quantification. 87% accuracy doesn't eliminate competitive balance when base-level data is universally available.

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube

Next: Soccer/Football AI revolution

#NBAnalytics #AIBasketball #SportsAnalytics #NBAtech #AsembleAI

143% growth for AI Engineers. 136% for Prompt Engineers. 135% for AI Content Creators. These aren't niches—they're fundamental new careers that couldn't exist before AI.

In this final "Who Survives the AI Shift" episode, Sam Dey and Mac Goswami reveal 16 brand-new job titles from 2025: Knowledge Architect, Orchestration Engineer, Conversation Designer, Human-AI Collaboration Leader. Top Emerging Roles: 

Prompt Engineer ($123K avg, top $200K+) - Building systematic AI outputs at scale. 40% fewer hallucinations, 60% better brand alignment.

AI Model Trainer - Fine-tune algorithms. Requires technical skills + deep industry knowledge.

AI Ethics Officer & Safety Analyst - Critical for governance in regulated industries. Assess biases, develop risk protocols.

Data Curator - Most accessible entry point. Domain expertise matters more than degrees.

Conversation Designer/NLP Engineer - Build chatbots, virtual assistants, translation systems.

AI Product Manager - Bridge technology and business with deep AI understanding.

AI Program/Project Manager - Handle AI implementation, operations, budgets. Huge growth projected. Where Jobs Are: 

Big Tech (Google, Microsoft, Amazon), AI-Native (OpenAI, Anthropic), Traditional Enterprises (JPMorgan, hospitals, retail) The Reality: 

New collar jobs exist at AI capability + human necessity intersection. Better AI needs MORE human oversight, not less. Consulting and freelancing booming—work that took days now takes hours.

The future belongs to those treating AI as collaborative tool, not competitive threat.

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Podbean

#AIJobs #PromptEngineer #FutureOfWork #AsembleAI

50% of employees need reskilling by 2026-RIGHT NOW. Are you ready, or already falling behind?

In this critical episode of "Who Survives the AI Shift," hosts Sam Dave and Mac Goswami expose the brutal reality: only 49% of employees feel equipped for their roles (down from 59% in 2024). Gen Z confidence crashed 20 points to 39%. The gap between awareness and action is where careers die. Key Takeaways: 

The Training Disconnect:

• 37% of employers claim they offer reskilling programs
• Only 28% of employees confirm these exist
• Companies check boxes without ensuring actual completion

Skills That Matter for 2030:

• AI & big data, cybersecurity, technological literacy
• Creative thinking, resilience, curiosity
• Winning combo: Technical fluency + human capabilities AI can't replicate

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube | Podbean

#Reskilling #AICareer #FutureProof #Upskilling #DataLiteracy #LifelongLearning #AsembleAI

Can AI amplify filmmaking creativity without killing the craft? Season 4 guest Sam Joos—20-year filmmaker, founder of AI Ad Studio and AI Film Society - shows how generative AI is transforming commercial production from $500K budgets to bedroom studios. Key Insights: 

The Breakthrough Moment: "Once I started prompting AI like I'd talk to a crew member on set, the cheat code unlocked." Sam went from AI skeptic to teaching 50+ filmmakers how to adapt.

The Economics Shift:

• Traditional commercials: $30K-$500K, 2-6 month turnarounds
• AI-powered: Shoot "London scenes" from home, deliver in 1-2 weeks
• Reality check: "It's not an easy button—taste and expertise still determine quality"

Quality vs. "AI Slop": What separates great AI work? Traditional filmmaking fundamentals-lighting, framing, camera movement, lens choice. "Hand a cinema camera to someone untrained—it'll look horrible. Same with AI tools."

Democratizing Film: Breaking Hollywood's gatekeeping: Midwest creators can now visualize ideas without industry connections, red carpets, or million-dollar budgets.

Your First Steps:

1. Study films/commercials you love—analyze what moves you
2. Learn cinematic vocabulary: shallow depth of field, steadicam, dolly shots
3. Research lighting, camera work, color grading techniques
4. Apply filmmaking knowledge to AI tools (MidJourney, Runway, Pika)
5. Build taste before prompts

Connect with Sam Joos:

🎬 AI Ad Studio 🎥 AI Film Society - Free resources, job boards, global community 📸 Instagram: @samjoosai

The Verdict: AI doesn't replace filmmakers, it creates AI-enhanced creators who blend craft with technology.

Subscribe: Spotify | Apple Podcasts | Amazon Music | iHeart Radio | YouTube

Which jobs are AI eliminating right now—not in five years, but today? In this hard-hitting episode of Inside AsembleAI, hosts Sam Dave and Mac Goswami examine the roles facing immediate AI displacement, backed by 2025 data showing actual job losses happening across industries. This is the episode nobody wants to hear but everyone needs to understand. What You'll Discover: 

Customer Service: The First Major Casualty

• 80% automation potential by 2025 (up from 60% recently)
• 2.8 million US customer service jobs at risk; 2.24 million likely displaced by 2025
• Real examples: Dukaan replaced 27 agents with ChatGPT bot, cut costs 99%, maintained 85% satisfaction
• IBM's AskHR handles 11.5M interactions annually with <5% human oversight, resolves 78% without escalation
• Why customers now prefer bots: 62% choose chatbots over waiting, 74% prefer bots for simple questions
• $8 billion in annual business savings driving rapid adoption

Data Entry: 7.5 Million Jobs on the Line

• Companies using AI form processing saw 56% reduction in data entry hiring rates
• Why it's vulnerable: quintessentially routine work—pattern matching, structured rules, accuracy-measured tasks
• AI eliminates human data quality issues while working faster and more consistently

Entry-Level White Collar Jobs: The Vanishing Career Ladder

• Anthropic CEO Dario Amodei's prediction: AI could eliminate half of entry-level white collar jobs within 5 years
• Entry-level marketing assistant roles dropped 31% since 2022
• Big Tech new graduate hiring down 25% (2024 vs 2023)
• Why entry-level specifically? Junior work = grunt work that AI now handles instantly
• The pipeline problem: eliminating training grounds that created pathways to senior positions

The Timeline Is NOW—Not Later:

• Salesforce cut 4,000 customer support roles (9,000 → 5,000)
• Sky Telecom eliminated 2,000 customer service jobs
• Microsoft laid off software engineers while CEO Satya Nadella revealed 30% of company code is now AI-written
• Displacement accelerating through 2027-2028

Critical Risk Factors for Your Job: ✓ Routine, predictable tasks ✓ Primarily data processing or pattern recognition ✓ Structured environments with consistent rules ✓ Cost savings dramatically outweigh human value-add

Who Bears the Biggest Risk:

• Southeast Asia: 52% increase in logistics/warehousing displacement since 2023
• Women: 9.6% at highest automation risk vs 3.2% for men (concentration in admin/customer service)
• Urban vs rural divide: 38% urban job postings include AI vs 14% rural

What You Should Do RIGHT NOW: Mac and Sam's urgent action plan:

1. Upskill toward AI-adjacent positions - learn to supervise, quality-check, and improve AI outputs
2. Transition to roles requiring human judgment - physical work, emotional intelligence, regulatory oversight
3. Pursue structural barriers - healthcare, skilled trades, positions AI can't easily automate
4. Don't wait - executives already rewarding employees who smartly implement AI into workflows

The Brutal Truth: If your tasks can be described in a detailed manual that someone could follow without judgment calls, AI can and likely will replace you. This isn't about being good at your job—it's about whether your job's fundamental nature aligns with AI's strengths.

Subscribe for the complete AI jobs series: YouTube, Spotify, Apple Podcasts, and Substack for in-depth articles.

Not all jobs are at risk from AI automation. In this episode of Inside AsembleAI, hosts Sam Dey and Mac Goswami reveal the safe zones—careers where AI enhances human work rather than eliminating it-and explain the crucial "why" behind these patterns so you can evaluate your own role's resilience. What You'll Learn: 

Healthcare: The Clearest Example of AI Augmentation

• 34 million new healthcare roles emerging by 2030 globally
• Nurse practitioners projected to grow 52% from 2023-2033
• AI healthcare spending rising from $15.1B to $19.8B, but it's augmenting, not replacing clinicians
• Why patients will always demand human faces for life-altering decisions—the trust factor AI can't overcome
• AI handles 15% (imaging, scheduling, protocols) while humans retain 85% (emotional support, complex diagnosis, ethical decisions)

The Four Traits of Automation-Resistant Careers:

1. Non-routine physical tasks in unstructured environments
2. Real-time sensory perception and 3D motor skills
3. Contextual problem-solving that can't be reduced to data
4. Human judgment under uncertainty and emotional complexity

Industries Where Humans Remain Essential:

Skilled Trades & Technical Work:

• Electricians, plumbers, construction workers face minimal AI threat
• Why troubleshooting a 100-year-old building requires detective work AI can't replicate
• 95% of skilled trade work demands hands-on human expertise navigating messy real-world constraints

Creative Leadership & Strategy:

• Brand directors, creative directors, strategic planners operating at psychology-culture-business intersection
• AI can draft content and analyze data (25% augmentation), but humans set vision and cultural direction
• Risk-taking, ethical accountability, and counter-cultural choices require human judgment
• Why AI struggles to navigate demographic sensitivities and cultural nuances in creative work

Education & Mentorship:

• Teachers won't be replaced because learning is fundamentally social
• AI tutors handle 20% (grading, practice, supplemental content)
• Humans retain 80% (inspiration, mentorship, emotional vs. intellectual struggle recognition)
• Special needs students, artistic children, and classroom dynamics demand emotional intelligence AI lacks

Your Career Action Plan: Sam and Mac provide practical guidance to audit your role:

• Identify automation risks: routine data processing, predictable patterns, structured environments
• Identify augmentation opportunities: human judgment, physical work, creative problem-solving, emotional intelligence
• Position yourself toward augmentation and embrace AI tools for routine tasks

The Bottom Line: Safe zones aren't static—they're determined by current AI capabilities and economic feasibility. As technology advances, new tasks requiring uniquely human skills will emerge. The jobs that remain safe provide value that's either technically impossible or economically impractical for AI to replicate.

Subscribe for More: Don't miss the next episode covering roles most vulnerable to AI automation. Subscribe on YouTube, Spotify, Apple Podcasts, and join our Substack for in-depth AI analysis.

Is AI really coming for your job? Or is the "AI apocalypse" just another tech scare story? In this episode of Inside AsembleAI, hosts Sam Dave and Mac Goswami cut through the fear-mongering headlines to examine what's actually happening in the AI job market right now - backed by hard data from the World Economic Forum, SHRM, Goldman Sachs, and Microsoft research. What You'll Discover: 

The Real Numbers Behind AI Displacement:

• 85 million jobs displaced by 2025—but 97 million NEW roles created (net gain of 12 million jobs globally)
• 23.2 million US jobs already 50%+ automated, yet 63.3% have barriers preventing complete replacement
• Why Microsoft's 200,000-user study shows AI is augmenting work, not eliminating it wholesale

Who's Actually at Risk:

• 58.87 million women vs. 48.62 million men in high-exposure roles—the demographic disparity nobody's discussing
• Why workers aged 18-24 are 129% more likely to fear job loss than those over 65
• How 49% of Gen Z believes AI has devalued their college education

The Historical Context:

• Why 85% of employment growth since 1940 came from tech-driven job creation, not destruction
• The pattern repeats: World Wide Web, cloud transition, and now AI—lessons from past transformations
• Goldman Sachs research: 0.3-point unemployment bumps are temporary, fading within two years

The New Jobs AI Is Creating:

• 350,000 emerging positions: Prompt engineers, AI ethics officers, human-AI collaboration specialists
• The catch: 77% require master's degrees—creating accessibility challenges for displaced workers
• Real examples from Microsoft, Cisco, Intel, and Meta layoffs vs. new AI role hiring

What This Means for YOU: Sam and Mac break down the transition vs. devastation reality—why this moment mirrors the World Wide Web revolution and cloud computing shift. You'll learn why pretending everything's fine OR catastrophizing about mass unemployment both miss the mark.

Subscribe for More AI Insights: Don't miss our next episode covering jobs AI will augment (not replace) and why those safe zones exist. Hit subscribe on YouTube, Spotify, or Apple Podcasts, and sign up for our Inside AsembleAI newsletter for weekly AI industry analysis.

Perfect for: Tech professionals, business leaders, career changers, students planning their future, and anyone wondering how AI will reshape work in the next five years.

AI analytics represents a fundamental shift from analyzing what happened to predicting what will happen. Traditional marketing analytics was retrospective-dashboards showing last month's performance, reports explaining why campaigns succeeded or failed. AI analytics is prospective-predictive models forecasting customer behavior, propensity scores indicating conversion likelihood, churn risk signals identifying at-risk customers before they leave.

The shift in marketing team composition is significant. Traditional teams were heavy on creative and campaign managers. AI-driven marketing teams need data scientists, analytics engineers, and marketing technologists who understand both strategy and technical implementation. The skillset evolves from "what message resonates" toward "what patterns in customer data predict behavior we can influence." 

Critical pitfalls include overfitting models on historical data, optimizing for proxies rather than actual business outcomes, and creating feedback loops where AI recommendations reinforce existing biases rather than discovering new opportunities. Privacy regulations like GDPR and CCPA create constraints on what data you can collect and how you can use it for profiling. 

The ROI is compelling. McKinsey research shows businesses using advanced analytics growing 10-15% faster than competitors, with 20-40% improvement in marketing efficiency through better targeting and resource allocation.

Servian Global Solutions projects that 95% of customer interactions will be AI-powered by 2025. We're in 2026 now-that's not a future prediction anymore, it's the present reality. The chatbot market is growing by $11.45 billion through 2026, fueled by major advances in natural language processing and machine learning making chatbots intuitive, context-aware, and capable of handling genuinely complex conversations.

Modern AI chatbots differ dramatically from frustrating automated systems of years ago. These systems now understand context, handle follow-up questions, detect sentiment, and maintain conversation flow naturally. They're not doing keyword matching scripts anymore—they're using transformer models similar to ChatGPT, trained specifically for customer service scenarios with reinforcement learning for real-time contextual awareness.

However, limitations exist. Chatbots struggle with truly novel situations they haven't been trained on, can't make judgment calls requiring human empathy, and occasionally hallucinate confidently incorrect information—which is why accuracy checking and clear escalation paths matter. Some customers simply prefer human interaction regardless of AI capability, which businesses must respect. 

Cost savings are substantial but shouldn't be the only driver. NIB Health Insurance saved $22 million through AI-driven digital assistance, reducing customer service costs by 60%. The strategic value extends beyond cost reduction: 24/7 availability supports customers globally, instant response times improve satisfaction, and consistent answer quality eliminates variance in agent knowledge.

Traditional ad buying involved manual targeting, static audiences, and fixed bids. AI advertising uses machine learning to optimize targeting, bidding, and creative selection in real time across millions of data points. Performance Max and Meta Advantage+ campaigns represent this evolution - algorithms handling what used to require entire teams of media buyers.

Smart bidding algorithms adjust bids based on conversion likelihood, time of day, device type, user behavior history, competitor activity, and dozens more variables simultaneously. This dynamic approach consistently outperforms manual bid management, especially for campaigns with large audiences and multiple ad variations. However, human strategy and oversight remain necessary—marketers must set clear goals, supply quality creative assets, and analyze performance to ensure AI automation aligns with business objectives.

Critical risks include over-optimization—AI might optimize for metrics that don't actually align with business goals. Optimizing for clicks gets clicks but might not deliver quality traffic. Optimizing for conversions without considering lifetime value might acquire expensive customers who churn quickly. The human role is defining success properly so AI optimizes toward meaningful outcomes.

Looking at 2026, programmatic advertising moves toward full automation. For small businesses without media buying expertise, this democratizes access to sophisticated advertising. For agencies and specialists, it forces evolution toward strategic consulting rather than tactical execution.

The numbers are staggering: 96% of companies now use generative AI for content production. Companies report 3-5x more content output, 30-50% cost savings, and 50% reductions in creation time. This isn't incremental improvement—it's transformational change in how marketing teams operate.

AI content creation in 2025 encompasses far more than ChatGPT writing blog posts. We're talking about integrated workflows governing ideation, creation, distribution, and analytics. Tools like Jasper, Copy.ai, and ContentBot handle everything from drafting to scheduling and multi-platform distribution. The sophistication has moved far beyond simple text generation.

Limitations remain clear: AI struggles with truly original creative thinking—breakthrough ideas that redefine categories. It excels at recombining existing concepts but genuine innovation requires human creativity. AI lacks emotional intelligence and cultural nuance, can mimic empathy but doesn't actually understand context the way humans do, and generates confidently wrong information (hallucinations), which is why human fact-checking remains non-negotiable.

Looking ahead, the strategic implication is marketing teams shifting focus from production to strategy. When AI handles volume, humans focus on insight, positioning, and differentiation. Small teams can now compete with large enterprises because production bottlenecks disappear.

AI personalization has evolved dramatically from basic segmentation to true individual-level customization. McKinsey's 2025 research shows businesses using advanced personalization techniques are seeing 10-15% revenue increases, with 89% of decision makers saying AI-driven personalization will be critical in the next three years. This isn't optional anymore-it's competitive survival.

Consumer expectations have shifted dramatically. 72% of consumers say they only engage with marketing messages tailored to their interests, and 90% are happy to share personal data if the result is a smoother, more personalized experience. However, they want immediate tangible value in exchange—brands can't just collect data and hope customers will be patient.

Looking ahead to 2026, generative AI will create not just personalized messages but personalized imagery, video, and even product configurations. Adobe's 2025 Digital Trends Report shows 58% of teams seeing GenAI ROI expect better quality customer interactions in the next 12-24 months. The winners will be brands that see personalization as a system, not just a tactic-building predictive models into planning cycles while maintaining human oversight on privacy and ethics.

Welcome to the final episode of the AI in Finance series, exploring algorithmic trading and AI market makers—genuinely the wild west of AI in finance. Here's context most people don't realize: 60-70% of equity market volume already comes from algorithmic trading, with high-frequency trading alone accounting for roughly 50%. When you think about the stock market, you're thinking about a system that's already majority AI and algorithms, not human traders.

Sam and Mac explore what fundamentally differentiates AI algorithmic trading from traditional algorithmic trading. Traditional algorithms follow fixed rules: if condition X, then execute action Y—deterministic and predictable. AI algorithms learn and adapt dynamically, recognizing complex patterns across multiple variables, adjusting strategies in real time based on changing market conditions, and optimizing behaviors continuously.

The technical models include reinforcement learning (AI learning optimal strategies through trial and error in simulations), LSTMs for time series prediction, and increasingly transformer models adapted for financial data—same basic architecture as ChatGPT but trained on market data instead of language. These models are exceptional at understanding that the same price movement means different things in different contexts: high volatility versus low volatility, bull market versus bear market.

Regulatory landscape remains challenging. The SEC requires reasonable oversight, but defining "reasonable" for systems executing thousands of trades per second is genuinely difficult. In practice, this means kill switches, risk limits built into algorithms, monitoring systems that flag unusual patterns, and automatic shutoffs when volatility triggers occur.

AI in credit decisions is genuinely controversial because it could either democratize lending and expand access to underserved populations or take historical discrimination and amplify it at scale. The reality is both are happening simultaneously in different institutions—it all depends on how intentionally the AI is designed and monitored for fairness.

Sam and Mac examine how AI is disrupting traditional credit scoring. FICO scores have dominated for decades using limited data: payment history, credit utilization, length of credit history, types of credit, and recent inquiries. This approach systematically excludes millions who don't have traditional credit histories, even if they're perfectly responsible with money and would be excellent borrowers.

The technical models include XGBoost as the industry standard and neural networks for processing more data with hidden layers. Traditional logistic regression is often a poor fit for real-world credit behavior. Banks need model governance with clear ownership, regular bias testing, robust explainability, and human oversight for complex cases. AI handles straightforward approvals and denials; humans handle the middle—complex situations requiring judgment and contextual understanding.

Compliance has traditionally been viewed as a pure cost center—regulatory overhead that doesn't generate revenue. But AI is fundamentally changing this equation by turning compliance from a defensive obligation into an actual strategic advantage. New LSTM networks are achieving 94.2% accuracy in compliance monitoring while simultaneously cutting false positives dramatically.

Sam and Mac explore why AI in compliance might be the biggest impact area that nobody is talking about. The false positive problem has always made compliance painful and expensive—traditional systems generated massive false positive rates, with analysts drowning in alerts where 95% turned out to be completely legitimate activity. This creates compliance fatigue where analysts become desensitized because so many alerts are false.

The episode covers AI's impact across major regulatory areas: AML (Anti-Money Laundering), KYC (Know Your Customer), Sanctions Screening, and Trade Surveillance. For AML, AI narrows down suspicious patterns while letting routine activity pass without alerts. For KYC, banks report 78% faster onboarding times and 85% reduction in manual review—customers approved in an hour instead of days.

AI must be transparent and auditable. The future is shifting from reacting to violations to preventing them entirely, flagging patterns on day three instead of catching problems on day 30, saving millions in potential federal lawsuits.

Over 50% of fraud now involves AI. FIDZY surveyed 562 fraud professionals globally and found AI-powered fraud has become the norm, not the exception. We're talking about deepfakes, synthetic identities, and AI-powered phishing so sophisticated it's basically indistinguishable from legitimate communications. The counter punch? 90% of banks are now using AI to fight back—fighting fire with fire.

Sam and Mac paint the threat landscape: deepfake calls that sound exactly like your bank's fraud department, using your bank's actual spoofed phone number, with perfect voice and professional script asking for your PIN. California bank customers received dozens of these calls and many fell for it because the technology is that convincing.

This is an arms race. Fraudsters use AI, banks use AI—there's no final victory. As bank AI gets smarter at detection, fraud AI evolves to evade those systems. It's like computer viruses and antivirus software—never-ending evolution and counter-evolution. The economic stakes are enormous: Deloitte estimates US banking losses from fraud could increase from $12.3 billion in 2023 to $40 billion by 2027, more than tripling in four years due to generative AI sophistication.

Human oversight remains essential. 88% of banking professionals say human oversight is non-negotiable. AI identifies potential issues and surfaces them to analysts, but humans make final calls on complex cases. The benefit: 43% of institutions report increased efficiency because AI handles high-volume straightforward cases, freeing human experts for complex nuanced cases requiring judgment.

Stanford just dropped a bombshell study: an AI analyst made 30 years of stock picks and outperformed 93% of human mutual fund managers by an average of 600 basis points—that's 6% annually. This is absolutely massive in the investment world, kicking off Inside AssembleAI's AI in Finance series with the technology that's shaking Wall Street.

Here's what's fascinating: the AI mostly used simple variables, not the sophisticated ones everyone expected. Firm size and dollar trading volume were dominant factors, but it used complex AI techniques to squeeze maximum predictive value from simple data everyone can access. The insight isn't about finding hidden data-it's about extracting more signal from obvious data. Any investment firm could have had this data in the pre-AI era, but it was simply too costly to justify economically.

Sam and Mac explore three main approaches institutions use today: pattern recognition for known scenarios (AI learns what fraud or manipulation looks like), anomaly detection for unknown threats (establishing what's normal and alerting on deviations), and predictive analytics for future behavior (forecasting what's likely to happen next). All happening in real time, in milliseconds-the game changer compared to legacy systems.

The data quality issue compounds everything—garbage in, garbage out. Models require at least five years of high-quality historical data for reliable results, and even then, past performance doesn't guarantee future success. Looking ahead to 2026, expect more hedge funds adopting sophisticated AI systems, models incorporating multi-modal data like satellite imagery and social sentiment, intensifying regulatory scrutiny, and continued democratization as retail investors gain access to tools that were hedge fund exclusive just years ago.