Stop Building AI Agents the Hard Way: Lessons from 25 Years in AI (w/ Rob Webster)

Is AI Really Taking Jobs? What's Going on Behind the Headlines

AI layoffs are dominating the news, but the story being told isn't the full story. In this episode, Jess and Kyle break down what's really driving the cuts, and what it means for your job and your business. When Meta announced 8,000 job cuts, the coverage landed the same way it always does: AI is replacing people, the future is here. But the numbers don't support that story. Meta's savings from the cuts amount to roughly three billion dollars. Their AI infrastructure spend this year runs to multiples of that. So what's actually going on? In this episode, Jess and Kyle work through the real economics behind the layoff headlines, from the infrastructure bets driving the cuts, to the compute costs that are now exceeding what companies spend on their own people, to the quietly alarming data on what's happening to early-career workers. They also cover the Musk v. Altman verdict, what it means for OpenAI's upcoming IPO, and why Anthropic keeps coming out looking like the adult in the room. The episode closes with practical guidance on what founders, team leaders, and employees can actually do right now, including why waiting to feel ready is the worst strategy available. What You'll Learn: * Why Meta's 8,000 job cuts are better understood as a budget-clearing exercise * What AI washing is and how to spot it in a layoff announcement * Why infrastructure spending at Amazon, Meta, and Microsoft is projected to exceed total payroll costs by $50 billion this year * What MIT research actually found when it tested whether AI is economically viable compared to keeping humans in the role * Why 43% of CEOs plan to reduce junior roles over the next two years * Why IBM's contrarian bet on junior hiring may look very smart in ten years * Why smaller businesses are better placed than large firms to make the same move * What AI fluency actually means in practice Tools we use and recommend: We only recommend tools we actually use. Both links below are affiliate links — if you sign up, it costs you nothing extra and helps support the show. Wispr Flow — AI-powered voice dictation that works across every app on your desktop - https://ref.wisprflow.ai/early-adoptr [https://ref.wisprflow.ai/early-adoptr] Granola — The best AI meeting notes! New users get 100% off for their first month. https://www.granola.ai?via=early-adoptr [https://www.granola.ai/?via=early-adoptr] Timestamps: 00:00 Introduction and Personal Updates 04:58 Why the AI Layoff Headlines Don't Tell the Whole Story 08:58 When Companies Cut Staff to Fund AI and Call It Efficiency 18:16 How Meta Extracted Its Employees' Knowledge Before Letting Them Go 22:40 Why the Productivity Gains Don't Justify the Scale of the Cuts 28:12 When Running AI Costs More Than Paying Your Team 29:58 MIT Research: AI Is Only the Cheaper Option in 23% of Cases 32:14 The Disconnect in AI Implementation 33:52 Why Junior Roles Are Being Cut First 35:44 The Talent Pipeline Problem Nobody Is Planning For 38:28 Redesigning Early-Career Roles Instead of Cutting Them 40:27 The Skills Gap in Education 42:29 Key Takeaways Resources: * https://www.shrm.org/topics-tools/news/technology/ai-layoffs-transformation-scapegoat [https://www.shrm.org/topics-tools/news/technology/ai-layoffs-transformation-scapegoat] * https://www.linkedin.com/news/story/ceos-plan-to-reduce-junior-roles-8108505/ [https://www.linkedin.com/news/story/ceos-plan-to-reduce-junior-roles-8108505/] * https://www.forbes.com/sites/danrunkevicius/2026/05/20/meta-layoffs-signal-ai-bill-is-coming-due/ [https://www.forbes.com/sites/danrunkevicius/2026/05/20/meta-layoffs-signal-ai-bill-is-coming-due/] * https://fortune.com/2026/03/31/marc-andreessen-ai-layoffs-silver-bullet-excuse-overhiring/ [https://fortune.com/2026/03/31/marc-andreessen-ai-layoffs-silver-bullet-excuse-overhiring/] * https://techcrunch.com/2026/05/19/openai-co-founder-andrej-karpathy-joins-anthropics-pre-training-team/ [https://techcrunch.com/2026/05/19/openai-co-founder-andrej-karpathy-joins-anthropics-pre-training-team/] * https://www.bbc.co.uk/news/articles/cewpyv79pw1o [https://www.bbc.co.uk/news/articles/cewpyv79pw1o] * https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html [https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html] * https://www.pymnts.com/artificial-intelligence-2/2026/jpmorgan-prioritizing-ai-hires-over-bankers/ [https://www.pymnts.com/artificial-intelligence-2/2026/jpmorgan-prioritizing-ai-hires-over-bankers/] * https://fortune.com/2026/05/22/microsoft-ai-cost-problem-early%20adoptr%20-agents/ [https://fortune.com/2026/05/22/microsoft-ai-cost-problem-early%20adoptr%20-agents/] * www.moneycontrol.com/technology/mark-zukerberg-s-leaked-viral-audio-clip-suggest-meta-is-tracking-employees-to-train-ai-article-13924715.html [https://airtable.com/appqH1fQ2nZDEMP6S/tblF8dtf6wNwkC5nS/www.moneycontrol.com/technology/mark-zukerberg-s-leaked-viral-audio-clip-suggest-meta-is-tracking-employees-to-train-ai-article-13924715.html?utm_source=www.theneurondaily.com&utm_medium=newsletter&utm_campaign=meta-used-staff-as-ai-training-data-then-cut-them&_bhlid=872adcab8f6bfed7ce96c095f227a025d327cd30&utm_source=www.theneurondaily.com&utm_medium=newsletter&utm_campaign=meta-used-staff-as-ai-training-data-then-cut-them&_bhlid=872adcab8f6bfed7ce96c095f227a025d327cd30] Get in touch with Early Adoptr: hello@earlyadoptr.ai [hello@earlyadoptr.ai] Follow Us on Socials & Resources: IG: https://instagram.com/early_adoptr [https://instagram.com/early_adoptr] TikTok: https://tiktok.com/@early_adoptr [https://tiktok.com/@early_adoptr] YouTube: https://www.youtube.com/@early_adoptr [https://www.youtube.com/@early_adoptr] Substack: https://substack.com/@earlyadoptrpod [https://substack.com/@earlyadoptrpod] ---------------------------------------- Hosted on Acast. See acast.com/privacy [https://acast.com/privacy] for more information.

Happy Birthday to Us: The AI Landscape Changed Completely in a Year, Here's What You Missed

A year in AI doesn't feel like a normal year, it feels like about five. In honour of Early Adoptr's first anniversary,  we thought it was worth stopping to take stock of what's changed in the last year. When we launched Early Adoptr, there was a pretty clear hierarchy, with ChatGPT was at the top, everyone else was catching up, and agents were something people talked about at conferences without anyone being entirely sure what they meant. Twelve months later, almost none of that is still true. In this episode, we take a proper look back at what actually changed. From the model landscape and the slow collapse of the "ChatGPT is the Google of AI" era, to reasoning becoming the baseline rather than the premium tier, to context windows going from a genuine operational headache to essentially a non-issue, the foundations shifted faster than most businesses could keep up with. We also get into what happened with agents once MCP — Model Context Protocol — gave them a shared language to work with, why agentic commerce looks completely different to how we predicted it would, what Answer Engine Optimisation means for any business that needs to be found online, and what has changed with the security picture once agents got access to real tools. We close out with a look ahead at what's actually worth paying attention to: outcome-based pricing, orchestrated multi-agent systems reaching smaller businesses, and what we're calling agent debt, the accumulating consequences of workflows that were built in a hurry and haven't been stress-tested yet. Thanks for being with us for the last year, and here's to the next 12 months! WHAT YOU'LL LEARN * Why reasoning models went from a premium add-on to the default , and what that shift enabled for agents and complex workflows * How context windows grew from a operational constraint to a non-issue, and what that unlocks for businesses working with large volumes of documents, contracts, or correspondence * Why smaller, more focused AI tools regularly outperform general-purpose models on the tasks they're built for, and what that means for how you structure your own stack * What MCP actually solved — and why it's the reason agents went from demo-quality to deployable for non-technical teams * What the two-tier internet looks like in practice and why it matters * Why ChatGPT's instant checkout failed commercially and what it tells us about how brands are learning to use AI for discovery * What AEO — Answer Engine Optimisation — means for any business that needs to be found online * How the security risk picture changed once agents got real access to real tools via MCP * What agent debt is * What outcome-based pricing means RESOURCES AND LINKS All previous episodes of Early Adoptr can be found here or via your podcast player of choice: https://shows.acast.com/early-adoptr [https://shows.acast.com/early-adoptr] Get in Touch: hello@earlyadoptr.ai [hello@earlyadoptr.ai] TikTok: @early_adoptr Instagram: @early_adoptr YouTube: @early_adoptr Timestamps: 00:00 What We've Been Up to This Week 04:14 One Year of Early Adoptr: What We Got Right (and Wrong) 06:27 The Model Landscape: Why ChatGPT Lost the Top Spot 11:01 AI Pricing Is Changing — and Your Bill Is Going Up 16:20 Context Windows: From Headache to Non-Issue 20:03 When Smaller Is Better: The Case for Specialist AI Tools 20:35 Use Cases for Small Language Models 22:53 gents: From Conference Buzzword to Actually Useful 29:02 What MCP Did for the Agent Ecosystem 31:29 Agentic Commerce and the Two-Tier Internet 37:55 How Brands Are Using AI for Discovery Without Losing the Customer 39:05 The Evolving Landscape of AI Security 44:17 The Shift in AI Risks and Management 48:26 From Subscriptions to Outcome-Based Pricing 50:22 AI Regulation, Memory, and the GDPR Question Nobody's Asking Yet 52:20 The Agent Debt Problem 56:12 Where Does AI Go From Here? Get in touch with Early Adoptr: hello@earlyadoptr.ai [hello@earlyadoptr.ai] Follow Us on Socials & Resources: IG: https://instagram.com/early_adoptr [https://instagram.com/early_adoptr] TikTok: https://tiktok.com/@early_adoptr [https://tiktok.com/@early_adoptr] YouTube: https://www.youtube.com/@early_adoptr [https://www.youtube.com/@early_adoptr] Substack: https://substack.com/@earlyadoptrpod [https://substack.com/@earlyadoptrpod] ---------------------------------------- Hosted on Acast. See acast.com/privacy [https://acast.com/privacy] for more information.

The Wrong Question: Why "How Do I Save Time With AI?" Isn't Enough

Most people start their AI journey by asking how to save time. That is not a wrong question — but Anthropic's latest research, based on open-ended interviews with over 81,000 Claude users across 159 countries, suggests it may not be the most important one. The most commonly reported productivity gain in the study was not speed. It was scope. Not doing existing work faster, but doing things that you simply couldn't before, because of budget, skills, or just the assumption that certain capabilities belonged to someone else. This episode is about the difference between saving time with the boring middle and asking what is now possible that wasn't before, and why that second question is where the real opportunity lies. What You'll Learn * Why the Anthropic study's methodology is unusual * The difference between efficiency gains and capability gains * How to identify your "boring middle" and what to do once you have sorted it * How to prevent your freed-up time from get absorbed back into more of the same * How a delivery driver and landscape gardener from the illustrate capability gains * What the Pocket OS incident reveals about AI agent permissions, and the simple rule that would have prevented it TRY GRANOLA If you've ever sat in a meeting, taken what felt like decent notes, and then opened them afterwards and they didn't capture anything, Granola is the tool for you. It runs in the background, captures everything, and turns your notes into something you can actually use. Both Jess and Kyle use it, it plays really nicely with Claude, and it is one of our most highly recommended tools New users get 100% off their first month using our link: granola.ai?via=early-adoptr [http://granola.ai?via=early-adoptr] Timestamps: 00:00 Introduction and Travel Plans 01:37 About the Anthropic Research 03:37 How the Study Actually Worked 07:36 The Headline Productivity Stat 10:35 The Four Types of Productivity Gain from AI 11:54 What the Data Says About Job Displacement 13:32 The Efficiency Game: What It Gets You and What It Misses 16:45 Why Automating the Wrong Things Makes You Faster at the Wrong Things 21:27 The Boring Middle: Why Consistency Is the Point 25:00 Capability Gains: Doing Things That Were Previously Off the Table 28:54 The Wrong Question: Efficiency vs. Capability 31:39 How Efficiency and Capability Feed Into Each Other 35:09 Practical Takeaways: What to Try This Week 38:09 AI News of the Week: Lessons from Pocket OS Incident Resources: * What 81,000 people told us about the economics of AI: https://www.anthropic.com/research/81k-economics [https://www.anthropic.com/research/81k-economics]  * https://www.linkedin.com/posts/oguzaliacar_anthropic-just-published-findings-from-80000-activity-7440462065959366657-IqJj/ [https://www.linkedin.com/posts/oguzaliacar_anthropic-just-published-findings-from-80000-activity-7440462065959366657-IqJj/]  * https://medium.com/activated-thinker/an-ai-interviewed-81-000-people-what-it-discovered-exposes-our-deepest-insecurities-c231c7d2f77e [https://medium.com/activated-thinker/an-ai-interviewed-81-000-people-what-it-discovered-exposes-our-deepest-insecurities-c231c7d2f77e]  * https://www.reddit.com/r/AI_Agents/comments/1ssz7v3/anthropic_surveyed_81000_claude_users_about_ais/ [https://www.reddit.com/r/AI_Agents/comments/1ssz7v3/anthropic_surveyed_81000_claude_users_about_ais/] Get in Touch: hello@earlyadoptr.ai [hello@earlyadoptr.ai] TikTok: @early_adoptr Instagram: @early_adoptr YouTube: @early_adoptr * What 81,000 people told us about the economics of AI: https://www.anthropic.com/research/81k-economics [https://www.anthropic.com/research/81k-economics]  * https://www.linkedin.com/posts/oguzaliacar_anthropic-just-published-findings-from-80000-activity-7440462065959366657-IqJj/ [https://www.linkedin.com/posts/oguzaliacar_anthropic-just-published-findings-from-80000-activity-7440462065959366657-IqJj/]  * https://medium.com/activated-thinker/an-ai-interviewed-81-000-people-what-it-discovered-exposes-our-deepest-insecurities-c231c7d2f77e [https://medium.com/activated-thinker/an-ai-interviewed-81-000-people-what-it-discovered-exposes-our-deepest-insecurities-c231c7d2f77e]  * https://www.reddit.com/r/AI_Agents/comments/1ssz7v3/anthropic_surveyed_81000_claude_users_about_ais/ [https://www.reddit.com/r/AI_Agents/comments/1ssz7v3/anthropic_surveyed_81000_claude_users_about_ais/] Get in touch with Early Adoptr: hello@earlyadoptr.ai [hello@earlyadoptr.ai] Follow Us on Socials & Resources: IG: https://instagram.com/early_adoptr [https://instagram.com/early_adoptr] TikTok: https://tiktok.com/@early_adoptr [https://tiktok.com/@early_adoptr] YouTube: https://www.youtube.com/@early_adoptr [https://www.youtube.com/@early_adoptr] Substack: https://substack.com/@earlyadoptrpod [https://substack.com/@earlyadoptrpod] ---------------------------------------- Hosted on Acast. See acast.com/privacy [https://acast.com/privacy] for more information.

What It Really Takes to Move Your Team Forward With AI (w/ Rob Webster)

This is the second part of our interview with Rob Webster, who has spent over 20 years in media and marketing, ran data and technology at MediaCom for some of the world's biggest brands, built and sold a MarTech and AdTech consultancy, and now works with enterprise businesses on how they actually adopt AI. In this episode, Jess and Kyle talk to Rob about everything from navigating the messy middle to what the future of work for juniors, where AI should play in your business, what it means for how you hire and develop people, why so many organisations are stuck between experimenting and scaling, and what it actually takes to move forward. We also cover the OpenAI vs Elon Musk trial, and how South Africa's AI Policy offers a useful reminder to always check your citations. What You'll Learn * How to identify your best AI use cases by starting with outcomes rather than tasks * Why AI multiplies what you're doing * How junior employees can move faster and take on more accountability earlier when they have AI as a working layer * How smaller businesses are now better placed to train entry-level hires than they've ever been. * Why real-world wisdom is the skill that can never be replaced. * What the messy middle of AI adoption looks like in practice and why most organisations are stuck in it * How leaders can model AI adoption in a way that actually moves teams forward TRY GRANOLA If you've ever sat in a meeting, taken what felt like decent notes, and then opened them afterwards and they didn't capture anything, Granola is the tool for you. It runs in the background, captures everything, and turns your notes into something you can actually use. Both Jess and Kyle use it, it plays really nicely with Claude, and it is one of our most highly recommended tools New users get 100% off their first month using our link: granola.ai?via=early-adoptr [http://granola.ai/?via=early-adoptr] Timestamps: 00:00 Intro & What We've Been Up To 04:09 The 10,000 Foot View: Why You Should Start Broad Before Picking an AI Use Case 10:42 What AI Means for Hiring and Training Junior Employees 14:55 Why Smaller Businesses Can Now Compete on Talent Development 16:20 Human in the Loop: Why Oversight Still Matters 19:35 The Messy Middle: Why Most Businesses Get Stuck Between Testing and Scaling 23:17 How Leaders Can Drive AI Adoption 34:02 AI News of the Week: OpenAI vs Elon Musk 34:31 AI Gone Wrong: South Africa's AI Policy Debacle 37:55 Wrapping Up for the Week Resources: * Rob Webster - https://www.linkedin.com/in/digitalstrategyleader/ [https://www.linkedin.com/in/digitalstrategyleader/] * TAU Marketing Solutions - https://taums.ai/ [https://taums.ai/] * AI Agents in Media Planning and Buying: https://taums.ai/ai-agents-in-media-planning-and-buying/ [https://taums.ai/ai-agents-in-media-planning-and-buying/] * Judging the Coming Wave of Agentic Adtech: https://taums.ai/judging-the-coming-wave-of-agentic-adtech/ [https://taums.ai/judging-the-coming-wave-of-agentic-adtech/] * Mastering the Messy Middle: https://taums.ai/mastering-the-messy-middle-with-a-balanced-approach-to-ai-implementation/ [https://taums.ai/mastering-the-messy-middle-with-a-balanced-approach-to-ai-implementation/] * 10 Predictions for Marketing and AI in 2026 (Futureweek): https://futureweek.com/rob-webster-10-predictions-for-marketing-and-ai-in-2026/ [https://futureweek.com/rob-webster-10-predictions-for-marketing-and-ai-in-2026/] * The AI Land Grab interview (AdWorldNews): https://www.adworldnews.com/news/robert-webster-tau-marketing-ai-land-grab [https://www.adworldnews.com/news/robert-webster-tau-marketing-ai-land-grab] GET IN TOUCH hello@earlyadoptr.ai [hello@earlyadoptr.ai] TikTok: @early_adoptr Instagram: @early_adoptr YouTube: @early_adoptr Get in touch with Early Adoptr: hello@earlyadoptr.ai [hello@earlyadoptr.ai] Follow Us on Socials & Resources: IG: https://instagram.com/early_adoptr [https://instagram.com/early_adoptr] TikTok: https://tiktok.com/@early_adoptr [https://tiktok.com/@early_adoptr] YouTube: https://www.youtube.com/@early_adoptr [https://www.youtube.com/@early_adoptr] Substack: https://substack.com/@earlyadoptrpod [https://substack.com/@earlyadoptrpod] ---------------------------------------- Hosted on Acast. See acast.com/privacy [https://acast.com/privacy] for more information.

Stop Building AI Agents the Hard Way: Lessons from 25 Years in AI (w/ Rob Webster)

It's here! The culmination of our series on agents, and if you've ever wondered how to make the most of the AI agents in your business (without a huge budget or a team of developers), this is the episode for you. This week on Early Adoptr, we are joined by Rob Webster, who has spent 25 years working at the intersection of data, machine learning, and marketing, including working on data and technology for brands like Dell, Tesco, and Coca-Cola. He now runs Tau Marketing Solutions, where he helps businesses adopt AI and build agents to solve real marketing problems. In this episode he joins Jess and Kyle and shares everything he has learned about making agents that actually work. From the "Fisher Price Agent" to building a daily action plan, the four components every working agent needs and why most agents fail, this episode is a goldmine of tips from years of experience. We also cover a major deal between SpaceX and Cursor, and what it tells us about where the real competition in AI is playing out right now. What You'll Learn * The two-prompt method Rob uses to turn a vague goal into a concrete daily action plan * The four components every working agent needs and the reason most agent setups fail to produce useful output * Why the most valuable skill in AI right now has nothing to do with technology, and how anyone can develop it * What human-in-the-loop looks like as a working habit rather than a safety concept * How to start with a "Fisher Price Agent" * Rob's tips for getting unstuck when you hit a wall TRY GRANOLA If you've ever sat in a meeting, taken what felt like decent notes, and then opened them afterwards and they didn't capture anything, Granola is the tool for you. It runs in the background, captures everything, and turns your notes into something you can actually use. Both Jess and Kyle use it, it plays really nicely with Claude, and it is one of our most highly recommended tools New users get 100% off their first month using our link: granola.ai?via=early-adoptr [http://granola.ai/?via=early-adoptr] Timestamps: 00:00 What We've Been Up to This Week 03:16 Interview with Robert Webster 05:07 AI News: SpaceX and Cursor Deal 05:47 Meet Rob Webster 10:51 The Two-Prompt Method: Going from Vague Goal to Concrete Plan 13:28 The Four Components Every Working Agent Needs 18:14 Why Knowing What Good Looks Like Is the Real Skill 20:26 Human-in-the-Loop in Practice 26:20 Building a Co-CEO Agent: From Fisher Price to Advanced 33:01 Where to Start If You Are Not Technical 37:42 Takeaways 40:07 AI News of the Week: SpaceX & Cursor Resources: * Rob Webster - https://www.linkedin.com/in/digitalstrategyleader/ [https://www.linkedin.com/in/digitalstrategyleader/ ] * TAU Marketing Solutions - https://taums.ai/ [https://taums.ai/] * AI Agents in Media Planning and Buying: https://taums.ai/ai-agents-in-media-planning-and-buying/ [https://taums.ai/ai-agents-in-media-planning-and-buying/] * Judging the Coming Wave of Agentic Adtech: https://taums.ai/judging-the-coming-wave-of-agentic-adtech/ [https://taums.ai/judging-the-coming-wave-of-agentic-adtech/] * Mastering the Messy Middle: https://taums.ai/mastering-the-messy-middle-with-a-balanced-approach-to-ai-implementation/ [https://taums.ai/mastering-the-messy-middle-with-a-balanced-approach-to-ai-implementation/] * 10 Predictions for Marketing and AI in 2026 (Futureweek): https://futureweek.com/rob-webster-10-predictions-for-marketing-and-ai-in-2026/ [https://futureweek.com/rob-webster-10-predictions-for-marketing-and-ai-in-2026/] * The AI Land Grab interview (AdWorldNews): https://www.adworldnews.com/news/robert-webster-tau-marketing-ai-land-grab [https://www.adworldnews.com/news/robert-webster-tau-marketing-ai-land-grab] GET IN TOUCH hello@earlyadoptr.ai [hello@earlyadoptr.ai] TikTok: @early_adoptr Instagram: @early_adoptr YouTube: @early_adoptr Get in touch with Early Adoptr: hello@earlyadoptr.ai [hello@earlyadoptr.ai] Follow Us on Socials & Resources: IG: https://instagram.com/early_adoptr [https://instagram.com/early_adoptr] TikTok: https://tiktok.com/@early_adoptr [https://tiktok.com/@early_adoptr] YouTube: https://www.youtube.com/@early_adoptr [https://www.youtube.com/@early_adoptr] Substack: https://substack.com/@earlyadoptrpod [https://substack.com/@earlyadoptrpod] ---------------------------------------- Hosted on Acast. See acast.com/privacy [https://acast.com/privacy] for more information.