Explore every episode of the podcast ctrl + alt + PR

Your website looks great. It just doesn't exist....at least not to the AI that's rapidly becoming how people find you.

Right now, there's a version of your site that humans see, and a completely different version that AI agents see when they crawl it. Most of the time, those two versions don't match. And most site owners have no idea.

Amber Krasinski and Erick Grau sit down with Kyle Duck, founder and CEO of Alli AI, to break down what an AI agent actually sees when it hits your website and why that gap between the human view and the machine view is quietly wrecking brand visibility, e-commerce accuracy, and even legal exposure.

For marketers and PR pros: if your content isn't machine-readable, you don't exist to the fastest-growing referral channel on the internet. This episode covers what's actually blocking you (or the brands you work with), why FAQs and plain language now beat old-school SEO tricks, and what agencies owe clients when visibility disappears overnight.

For the tech and IT crowd: We get into the mechanics — JavaScript rendering, pre-rendering engines, the real difference between crawlers/scrapers/agents, and Cloudflare's new default AI-crawler blocking rules taking effect September 15, 2026, which could silently cut off a new site's visibility from day one.

What We Cover:

• Why AI crawlers see raw HTML, not what loads in your browser and what that means for JavaScript-heavy sites
• A live demo where Kyle pulls up a major website and shows exactly what ChatGPT can and can't see
• The Fortune 500 visibility study: 250 sites, 20,000 pages, 27% of content invisible
• Why default Lovable and vibe-coded sites are 0% visible to ChatGPT out of the box
• Cloudflare's September 15, 2026 default crawler-blocking change and what it means for new sites
• The EU vs. US regulatory gap on AI misinformation liability, including a recent German court ruling against Google Gemini
• Why Kyle is more bullish on the future of agencies and consultants than he was five years ago

Key Takeaways:

• If AI can't see your site, you don't get cited...and if you do get cited, it's more likely to hallucinate answers.
• Content fully accessible to AI is cited significantly more often.
• Agencies managing client sites need AI visibility audits as standard practice.
• Regulatory pressure is emerging in the EU first; expect US businesses to follow for compliance reasons.
• Easy to use site builders and vibe-coded tools (Lovable and similar) are invisible to AI by default, so check before you launch!

Guest: Kyle Duck, Founder & CEO, Alli AI — alliai.com | LinkedIn: Kyle Duck | X: @theKyleDuck

🔔 Subscribe so you don't miss future episodes exploring the worlds of emerging tech, marketing, PR, and cybersecurity.

🌐 Visit us at ctrlaltpr.com

📲 Follow Ctrl + Alt + PR on LinkedIn and Instagram @ctrlalt.pr

Every week there's a new "must-have" AI agent, and every week somebody spends half a day setting it up before the next one drops. This episode's about knowing which shiny objects are actually worth looking into.

Amber and Erick tear into the agent gold rush (Grok Bot, the open-source Buzz framework, DeepSeek, and running models locally through Ollama) and get real about the difference between the harness (the app you're typing into) and the LLM (the actual model doing the thinking).

They also get into why "trust the black box" isn't a strategy, what actually causes AI hallucinations, and why the smartest move right now might be doing less, not more.

Things we cover for marketers and PR pros: the brand-risk math of feeding proprietary info into a chief-of-staff agent, why "it's the harness, not the model" changes vendor-lock conversations, and how to evaluate a shiny new AI tool without the FOMO tax.

Things we cover for techies: harness vs. LLM architecture, self-hosting local models via Ollama, open-source multi-agent orchestration (Buzz), the real cause of hallucinations (probability, not intent), and Erick's actual RAM/hardware math for running models on your own machine.

• Harness vs. LLM: the one distinction that explains almost every AI vendor-lock problem
• Grok Bot's chief-of-staff/engineer agent model — legit or just well-marketed hype
• Buzz, the open-source multi-agent framework built to run entirely on hardware you control
• Running models locally via Ollama, and the real RAM/hardware tradeoffs of doing it
• Why hallucinations happen (it's math, not malice) and how to build a verification habit around it
• The Commerce Department's export-control pause on Anthropic's Fable/Mythos models, and what it means for access to frontier AI
• Turning a messy, one-off AI workflow into a reusable, shareable skill your whole team can use

• The harness is the risk, not the model. Know which one you're actually trusting with your data.
• Hallucinations aren't a bug in judgment — they're probability doing exactly what it's built to do. Verify anyway.
• Don't chase every new agent release. If it's actually good, you'll hear about it in three weeks and it'll be more stable by then.
• Open source isn't a vibe, it's an audit trail — you can see exactly what the code does before you trust it with anything.
• The ROI question isn't "can AI do this," it's "is this worth the setup time given how fast the tooling is still moving."

Tools Mentioned: Grok Bot · Buzz · DeepSeek · Ollama · Claude · ChatGPT · Qwen · Kimi · Gemma

🔔 Subscribe so you don't miss future episodes exploring the worlds of emerging tech, marketing, PR, and cybersecurity.

🌐 Visit us at ctrlaltpr.com

📲 Follow Ctrl + Alt + PR on LinkedIn and Instagram @ctrlalt.pr

A predictive model isn't useful because it's fast. It's useful because it actually tells you why something happened.

In this episode, Amber and Erick sit down with Evante Daniels, founder of SEEQER, a predictive behavioral research firm, to unpack how his team traces consumer and cultural trends back to their real root cause instead of just reporting that they're happening.

Evante breaks down the dangers of using data models that are still working off a pre-2020 baseline, what generation of consumers is statistically diverging from all the others, and how root-cause thinking changes the way a brand's teams respond when a situation starts to turn.

The conversation also explores ambient stress tracking, media literacy in an algorithm-driven world, data-driven brand marketing tactics, and why "trusted" sources aren't always the same as accurate ones.

Things we cover for marketers and PR pros: root-cause data vs. trend reporting, how to nonprofit marketing can inform tactics for big brands, why trends happen, auditing your data sources, and how to reach niche, trusted audiences

Things we cover for techies: how SEEQER models data as "macroeconomic triangulation" across industry sensitivity, geography, and ethnic group rather than standard trend-scraping, the self-hosted stack behind their engine, and Evante's real-time take on shifting research workloads off certain hosted LLMs

Guest: Evante Daniels

Website: seeqer.co

What We Cover

• How SEEQER traces consumer and cultural trends back to their actual root cause, not just the moment they went viral
• Why data models built before 2020 are quietly misleading leadership decisions today
• The Gen Alpha and Gen Beta discontinuity, and why these generations don't fit marketing assumptions
• What "ambient stress tracking" is and how it shapes consumer and employee behavior
• Evante's actual media literacy routine for separating signal from institutional and algorithmic noise
• How root-cause data changes the way a team responds when a narrative starts to shift

Key Takeaways

• A trend on a dashboard tells you what happened. It doesn't tell you why — and why is what actually informs a decision.
• If your data baseline predates 2020, you're not looking at "normal." You're looking at a different world wearing the same label.
• Gen Alpha isn't a smaller version of Gen Z. Their relationship to risk, stability, and institutions is a real statistical break from every generation before them.
• Media literacy isn't a checklist. It's a repeated habit of consuming multiple sources and asking "where did you hear that?" instead of "how could you believe that?!?"
• Knowing the root cause of a shift is worth more than reacting fast to the symptoms of it.

Tools Mentioned: ChatGPT · Claude · DeepSeek · SEEQER (Q) · Docker · Postgres

🔔 Subscribe so you don't miss future episodes exploring the worlds of emerging tech, marketing, PR, and cybersecurity.

🌐 Visit us at ctrlaltpr.com

📲 Follow Ctrl + Alt + PR on LinkedIn and Instagram @ctrlalt.pr

A public relations program isn't something you build out once. It's an operating system you never stop updating.

In this episode, Amber and Erick sit down with Gini Dietrich, founder of Spin Sucks and creator of the PESO® Model, to unpack what actually changes when AI enters communications and what a decade-old prediction about PR got right.

Gini breaks down why the PESO® Model evolved from a static framework into a full operating system, why "AI doesn't work" is almost always a prompting problem, and what happens when a comms team lets AI draft a crisis statement without adding real judgment to it.

The conversation covers visibility engineering, the "trust recession," why most executives still don't understand what a communications team actually does, and what's really at stake when a brand's holding statement reads like every other AI-written holding statement.

Things we cover for marketers and PR pros: how to audit AI outputs for accuracy, the PESO® Model as an operating system, AI-drafted crisis statements, integrated marketing tactics, visibility engineering, brand trust erosion, and why leadership keeps asking if they still need a comms team.

Guest: Gini Dietrich

Website: spinsucks.com

What We Cover

• How Gini predicted PR's evolution over a decade ago and what she got wrong about the timeline
• The difference between a framework you learn once and an operating system that adapts
• Why "AI doesn't work" is almost always a prompting problem, not a tool problem
• How AI-drafted communications can (and should) be improved
• What visibility engineering actually means and why it matters for PR pros
• What happens when AI content erodes credibility at scale
  

Key Takeaways

• AI doesn't replace communications judgment. It replaces the busywork around it.
• A framework you learn once and stop thinking about is already outdated. To keep up, you have to adapt.
• The PESO® Model still starts with Owned and Earned — trust has to exist before Paid amplifies.
• Most companies asking "do we still need PR" don't understand what PR does in the first place.
• AI-generated content that mirrors every other AI-generated content is how brands lose trust fastest.
  

Tools Mentioned: ChatGPT · ChatGPT Voice Mode · NotebookLM · Claude

🔔 Subscribe so you don't miss future episodes exploring the worlds of AI, cybersecurity, PR, media, and emerging tech.

🌐 Visit us at ctrlaltpr.com

📲 Follow Ctrl + Alt + PR on LinkedIn and Instagram @ctrlalt.pr

AI video isn't coming. It's already rewriting production, marketing, and creative work in real time.

In this episode, Amber and Erick sit down with Bob Demello, VP, Creative Director of Video at Relevate Health, to unpack what happens when generative video tools become good enough to blur the line between production workflow and synthetic reality.

Bob breaks down how AI is already cutting production costs, accelerating editing workflows, reshaping commercial video, and forcing brands to rethink trust, authenticity, and creative labor. The conversation dives into AI-generated actors, synthetic media, deepfake fears, pharma compliance chaos, and why most companies still have no idea how to govern this technology internally.

Things we cover for marketers and PR pros: AI-generated spokespeople, brand trust, crisis response, synthetic media disclosure, and how audiences react when “real” becomes negotiable.

Things we cover for creatives and video teams: AI production workflows, prompt-driven directing, and why AI still doesn’t replace taste, experience, or storytelling instincts.

Guest: Bob Demello

Website: www.relevatehealth.com

What We Cover

• Why AI video production costs are collapsing
• The difference between AI as a production tool vs deceptive media
• Why pharma and healthcare brands are panicking about AI governance
• How tools like Nano Banana are changing video editing
• AI-generated actors, avatars, and synthetic presenters
• Why prompting is becoming a new creative discipline
• The fear creatives have around job security and automation
• Why “spotting fake AI” is becoming impossible fast

Key Takeaways

• AI doesn’t magically make good content. It amplifies the people who already understand storytelling, production, lighting, editing, and direction.
• Most brands are still making AI policy decisions from a place of fear, not understanding.
• The ethical problem isn’t the tool itself. It’s consent, disclosure, and audience trust.
• AI video workflows are already reshaping marketing, healthcare communications, and commercial production.
• Creative professionals who learn AI tools now will have a massive advantage over teams pretending this shift isn’t happening.
• People keep saying AI replaces creativity while generating the visual equivalent of expired LinkedIn motivational posters. Grim stuff.

Tools Mentioned: Sora · Kling · HeyGen · Synthesia · Nano Banana · ChatGPT · Midjourney · Flux · Creative Ref

🔔 Subscribe so you don’t miss future episodes exploring the worlds of AI, cybersecurity, PR, media, and emerging tech.

🌐 Visit us at ctrlaltpr.com

📲 Follow Ctrl + Alt + PR on LinkedIn and Instagram @ctrlalt.pr

Deepfakes aren't just a strange concept. They're a present-day brand liability, a cybersecurity crisis, and a PR nightmare waiting to happen.

In this episode, Amber and Erick sit down with Josh McKenty, CEO of PolyGuard AI and co-founder of OpenStack, to get tactical about synthetic media, AI fraud, content, and digital identity verification.

Josh explains why every deepfake detection tool is already losing the arms race, breaks down the C2PA provenance standard Adobe and Microsoft are quietly building into the content ecosystem, and makes the case that impersonation fraud (not Hollywood-level fakes) is the attack vector quietly devastating businesses right now.

Things we cover for marketers and PR pros: AI-generated spokespeople, synthetic influencers, brand authenticity, and crisis response in the first 60 minutes of a deepfake or AI clone-related issue

Things we cover for IT and security leaders: the DPRK IT worker scheme and how enterprise identity verification is evolving fast

Guest: Josh McKenty, CEO of PolyGuard AI

Website: polyguard.ai

What We Cover

• Deepfake vs. AI clone — and why the distinction matters legally
• Why GANs make detection a permanent losing game
• The C2PA standard and what it means for brands and journalists
• What PolyGuard actually does (hint: not detection)
• The DPRK IT worker scheme — fun fact: Google once had 50 North Korean spies on payroll...oops
• AI-generated spokespeople: ethical line or legal liability?
• The virtual kidnapping scam and the one move that stops it
• Section 230, DMCA, and a policy fix that doesn't ban any tools
• Deepfake crisis comms — what to do in the first 60 minutes

Key Takeaways

• Detection is a false sense of security. The tells (extra fingers, metallic voices, lighting errors) are fading FAST and have been for years.
• Using someone's likeness without consent isn't just an ethics problem. It's theft of endorsement, and the legal exposure is real.
• A deepfake crisis is a reputation problem first, an IT problem second. Silence reads as guilt.
• Build your credibility anchors now (C2PA signing, a known domain, a verified presence) before you need them.
• Social platforms have no real incentive to protect your brand. Your response protocol can't depend on them moving fast.
• A deepfake doesn't have to be convincing to everyone. It just has to be convincing to the right audience at the right moment.

Tips from Josh

• Stop answering unknown calls. Set your phone to ring only for contacts.
• Hang up on loved ones calling from strange numbers, then call them back immediately. It feels wrong. That's the point.
• Turn on MFA everywhere, especially email. It's the master key to everything else.
• Stop trying to spot fakes. That skill no longer exists and the tech's getting really good, really fast.

Tools Mentioned: PolyGuard AI · C2PA · TruePic · HeyGen · ElevenLabs · IC3 · Verizon DBIR · Signal

🔔 Subscribe so you don't miss an episode.

🌐 Visit us at ctrlaltpr.com

📲 Follow us on LinkedIn at Ctrl + Alt + PR and on Instagram @ctrlalt.pr

Welcome to Ctrl+Alt+PR, the show where communications meets technology (and neither side comes out unscathed.)

Every episode, Amber Krasinski (founder, media literacy nerd, and comms pro) and Erick Grau (founder, CEO, and AI early adopter) pull back the curtain on the conversations that PR, marketing, and tech people should already be having. Deepfakes. AI fraud. Digital identity. Brand visibility. Crisis comms. You name it.

Basically all the stuff keeping your CISO and your CMO up at night, but finally in the same room.

No jargon left unexplained. No spin left unchallenged. No tool left unexplored.

Stick around because things are about to get interesting!

🔔 Subscribe so you don't miss an episode.
🌐 Visit us at ctrlaltpr.com
📲 Follow us on LinkedIn and Instagram