In The Loop – Détails, épisodes et analyse

Are we witnessing the end of entry-level jobs as we know them? In this episode of In The Loop, I’m tuning in from a shed on the coast of Scotland—yes, seriously—to dig into one of the most urgent and overlooked consequences of AI: its impact on early career opportunities. 

Based on fresh insights from SignalFire’s State of Tech Talent Report 2025, World Economic Forum, and LinkedIn data, I explore how AI and automation are rapidly dismantling traditional junior roles and what this means for economic mobility, education, and the next generation of talent.

I’ll unpack the widening disconnect between digital labor and human development, the rise of experience paradoxes, and whether employers have a structural responsibility to train and invest in new talent. If you're a graduate, a hiring manager, or simply concerned about the future of work, this is one episode you can’t afford to miss.

⏩ Episode Highlights

(00:15) - The State of Talent Report 2025: the urgent questions raised by tracking 650 million professionals’ career prospects.

(02:15) - The data: a 50% drop in junior hiring and the “experience paradox”—is AI creating a Lost Generation?

(06:00) - The societal impact: rethinking education in a world with fewer traditional pathways—and the moralities. 

(08:00) - Commercial responsibilities: global apprenticeship models that might offer solutions.

(11:25) - The call to action: Why 2025 might be the inflection point we can’t ignore.

🔗 Links & Resources

📊 SignalFire’s State of Tech Talent Report 2025

📖 Read the written version of this episode on the Mindset AI blog (with all relevant links and resources). Read here.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

In this episode of In the Loop, I dive into the biggest AI reveals from Microsoft’s Build and Google I/O—two landmark conferences that showcased over 150 new AI features in just one week. But this isn't just a recap. I cherry-pick the announcements that I believe signal the future of AI, particularly in how Microsoft and Google are taking radically different routes in shaping our digital world.

From Microsoft’s bold vision of an "open agentic web" to Google’s consumer-driven experiments and AR glasses revival, I break down what these strategies mean for developers, businesses, and everyday users. Whether you're an enterprise builder, a curious consumer, or just someone trying to keep up with the rapid pace of AI innovation—this one’s for you.

⏩ Episode Highlights

(00:40) — 7 days, 150 AI features, and AI agents mentioned 300 times: an overview of Microsoft Build and Google I/O 2025

(01:23) — Microsoft Build: Open Agentic Web, Model Context Protocol, and layoffs → Enterprise AI strategy

(07:10) — Google I/O: Agent Mode, AI Search & Shopping, wearables → Consumer AI strategy

🎧 Related Episodes

What is Model Context Protocols (MCP)? At 3:55

What is Agent2Agent Protocol (A2A)?

🔗 Links & Resources

Microsoft Build overview

Google I/O 2025 overview

📖 Read the written version of this episode on the Mindset AI blog (with all relevant links and resources). Read here.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

The world is buzzing about Model Context Protocol (MCP)—but why does it matter so much? In this episode of In The Loop, I break down what MCP is, why it’s being compared to the protocols that shaped the internet, and why it could be the key to unlocking the next wave of AI agent adoption. If you've ever wondered how AI agents can seamlessly connect to real-world applications—without endless custom integrations—MCP might just be the answer.

We’ll dive into the historical parallels between MCP and protocols like HTTP and SMTP, explore the frustrating bottlenecks that have limited AI agent capabilities, and discuss how a standardized framework could lead to an explosion in AI innovation. Whether you're an AI developer, tech enthusiast, or just curious about the future of AI, this episode will give you a clear, no-nonsense breakdown of why MCP could be the game-changer we’ve been waiting for.

Episode Highlights

(1:10) – What are protocols, and why have they been essential in shaping the internet?

(4:30) – Enter MCP: Protocols in the context of AI and LLMs.

(7:55) – The definition: What is MCP for AI agents?

(12:55) – Closing thoughts: The rapid evolution of AI agent capabilities—what happens when MCP is widely adopted?

Links & Resources:

📖 Read the written version of this episode on the Mindset AI blog (with all relevant links and resources). Read here.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

The way we build software is changing—fast. This week, Dario Amodei, CEO of Anthropic, made a bold prediction: within six months, AI will write 90% of all code. Meanwhile, “vibe coding” has become one of the hottest topics in AI, even surpassing "prompt engineering" in Google searches. And then there’s Cursor, an AI-powered coding startup that has skyrocketed by 9,900% in just a year, hitting $100 million in revenue. But what does all of this mean for developers? Will traditional programming be replaced, or are we witnessing yet another shift in how software is created?

In this episode, we dive into Amodei’s claim, the rapid rise of vibe coding, and the broader implications for businesses, developers, and the tech industry as a whole. We’ll also explore historical parallels—how each wave of technological automation has transformed rather than eliminated roles—and discuss what the next generation of software development might look like.

Episode Breakdown:

(1:10) – Dario Amodei’s bold claim: AI will write 90% of code in six months. ▶️ Watch the YouTube Short here.

(5:00) – The explosive rise of vibe coding—what it is and why it's trending.

(10:05) – Tim O'Reilly's take on the ever-changing role of programmers.

(14:00) – Historical patterns: How automation always changes the trajectory of jobs. 🎧 Listen to the mentioned podcast episode about Satya Nadella's vision for AGI.

(16:10) – What’s next? The new frontier of programming and problem-solving. ▶️ ⁠Watch the full YouTube video of children's predictions for the future here.⁠

Links & Resources:

📖 Read the written version of this episode on the Mindset AI blog (with all relevant links and resources). Read more here.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

What if AI agents started making decisions for us—and we didn’t even realize it? 

In this episode of In The Loop with Jack Houghton, we dive into the viral “gibberlink” moment—when two AI agents seemingly created their own secret language—and why it’s more hype than reality. But beneath the noise, a quiet revolution is happening: multi-agent systems. These specialized AI agents collaborate, communicate, and tackle complex tasks far more efficiently than a single model ever could—and they’re about to reshape business, technology, and culture in ways we’re only beginning to understand.

We’ll break down what multi-agent systems really are, how they work, and the opportunities (and challenges) they bring. From autonomous workflows and digital assistants to the rise of business-to-agent (B2A) services, this is the next big leap in AI evolution. Stick around to hear why this quiet revolution is set to explode over the next 18 months—and what it means for your life and work.

Episode Highlights:

(01:15) — The viral “gibberlink” video: What really happened between two AI agents?

(06:20) — What are multi-agent systems and how do they work?

(10:10) — The challenges of multi-agent systems: Scalability, security, and coordination

(17:35) — Why business-to-agent (B2A) services are the next big tech shift

(20:40) — How multi-agent AI could reshape business and our entity society.

Links & Resources:

📖 Read the written version of this episode on the Mindset AI blog (with all relevant links and resources). Read here.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

The world of software and technology is undergoing a massive shift—and AI agents are at the center of this revolution. In this episode of In The Loop, Jack Houghton dives into how vertical AI agents, like OpenAI’s deep research agent and Perplexity, are changing the game by consolidating complex workflows into single, powerful agents. Imagine replacing an entire stack of SaaS tools with one AI-driven system. It’s not just possible—it’s already happening.

We explore how these multi-agent systems work, why they’re so disruptive to established players like Google, and how they’re reshaping industries like healthcare, law, and enterprise software. Plus, Jack shares insights on how SaaS companies can adapt (or risk getting left behind). If you want to understand where AI is taking the future of work and tech, this is an episode you don’t want to miss.

Episode Highlights:

(01:10) — The rise of vertical AI agents and why they threaten traditional SaaS models

(04:25) — Why Google’s business model is at risk from AI-driven search tools

(08:35) — Why SaaS companies must embrace AI agents or face being left behind

(10:00) — How vertical AI solutions could entirely unbundle today’s SaaS landscape

Links & Resources:

📖 Read the written version of this episode on the Mindset AI blog (with all relevant links and resources). Read here.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

In this episode, I reflect on an incredible conversation between Satya Nadella, CEO of Microsoft, and Dwarkesh Patel on the Dwarkesh Podcast—and trust me, there’s a lot to unpack. Nadella offers a refreshing and sometimes critical take on the current AI boom, challenging our obsession with AGI (Artificial General Intelligence) as the ultimate benchmark of success.

I dig into some big-picture questions: Is AI really on the brink of a new Industrial Revolution, or is that an overstatement? What’s the true measure of AI’s impact: human-level cognition or real-world economic growth? And how do we ensure these breakthroughs benefit everyone, not just a select few? This episode explores why these questions matter and what they reveal about the future of work, innovation, and society.

Episode Highlights:

(01:25) — Setting the scene: about the Dwarkesh Podcast’s latest episode with  Satya Nadella

(03:40) — Satya Nadella’s take on why AI’s Industrial Revolution comparison might be off

(05:25) — The difference between knowledge work and knowledge workers, and why AI won’t replace human creativity

(07:20) — The “10x human” concept and why AI should amplify, not replace human potential

(08:50) — Why AI needs to reshape workflows and how companies keep getting it wrong

(10:45) — What lean manufacturing teaches us about AI’s potential in the workplace

(13:25) — The regulatory and social barriers we must address for widespread AI adoption

(14:30) — Why Satya Nadella calls AGI a “nonsensical benchmark” and what we should focus on instead

Links & Resources

📖 Read the written version of this episode on the Mindset AI blog (with all relevant links and resources). Read here.

🎧 Listen to the Dwarkesh Podcast on Spotify.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

We tend to think that technological progress is inevitable. But history tells a different story—one where innovation isn’t just about power but about accessibility. Just as computers needed intuitive interfaces to move from clunky calculators to the personal devices we use today, AI now faces a similar roadblock. The challenge isn't just building more powerful models—it's designing experiences that make AI simple, intuitive, and trustworthy.

In this episode, I take a deep dive into why this problem is so important (and so hard to solve). We’ll go back to the 1968 Mother of All Demos, explore how Xerox PARC and Apple redefined computing, and connect those lessons to today’s AI landscape. Why do most people still struggle to trust AI? What are the biggest hurdles in human-agent communication? And most importantly—what needs to happen next? Let’s break it all down.

Episode Highlights

(01:40) - Why AI isn’t just about impact—it’s about usability and trust

(03:50) - How Douglas Engelbart’s 1968 demo changed computing forever

(07:20) - The Xerox PARC innovations that Steve Jobs saw—and simplified

(09:25) - The biggest UX challenges holding AI agents back

(14:00) - Popular agent design frameworks leading to the future

Links & Resources

📖 Read the blog version of this episode on the Mindset AI blog (with all relevant links and resources). Read here.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

AI agents are everywhere—but what do they actually do?

If you thought AI agents were about to run your life, automate every task, and independently manage your business… you might be disappointed. But make no mistake—agentic AI is the future.

In this episode, I cut through the hype and break it all down in plain, no-nonsense terms:

✔️ What AI agents really are (and what they’re not)

✔️ How they work—including reasoning, memory, and tool use

✔️ The difference between horizontal, vertical & narrow agents

✔️ How OpenAI’s Operator & Deep Researcher stack up

✔️ Where AI agents are headed & why they still need humans

By the end of this episode, you’ll have a crystal-clear understanding of agentic AI—without the marketing fluff or unrealistic promises.  

Episode Breakdown

(00:55) – What is an AI agent? Breaking through the buzzwords

(02:10) – The difference between agentic AI, workflows & automation

(06:45) – OpenAI’s Operator & Deep Researcher: Game-changers or just hype?

(13:45) – The future of AI agents: What’s next & who stands to win?  

Links & Resources

📖 Read the blog version of this episode on the Mindset AI blog (with all relevant links and resources). Read here.

If you enjoyed this episode, rate, follow, and share! It helps others stay ahead of the latest AI trends. 🚀

DeepSeek just shook up the AI world. They’ve released an open-source AI model that—on paper—rivals the best from OpenAI and Google. But here’s the kicker: they claim to have trained it for just $5.6 million—a fraction of what U.S. labs spend. Even crazier, inference costs are so low that some developers are running 200,000 API calls for just $50.

If this is true, it’s not just a big deal—it’s a threat to the entire AI industry as we know it. In today’s episode, I break it all down in plain English:

✔️ What DeepSeek actually did (and how they made AI faster, smaller, and cheaper)

✔️ Why it’s causing panic in Silicon Valley

✔️ The real impact on AI companies, Nvidia, and the open-source movement

✔️ Is this a true paradigm shift or just another AI hype cycle?

By the end of this episode, you’ll have a crystal-clear understanding of what’s really going on—without the jargon, the hype, or the noise.

Episode Breakdown

(01:00) – What is DeepSeek, and why is everyone talking about it?

(02:20) – The four major breakthroughs that make DeepSeek different

(07:45) – The geopolitical and business implications—who wins, who loses?

(12:45) – My take: Is the market reaction justified, or is this overblown?Links & Resources

📖 Read the blog version of this episode on the Mindset AI blog (with all relevant links and resources). Read now.

If you found this episode insightful, don’t forget to rate, follow, and share! It helps more people stay ahead of the latest AI breakthroughs. 🚀