In partnership with

We track Product so you don't have to. Top Podcasts summarised, the latest AI tools, plus research and news in a 5 min digest.

Hey Product Fans!

Happy Easter 🐰 Welcome to this week’s 🌮 Product Tapas.

New here? We're your shortcut to staying sharp. Essential stories, practical tools, real insights.

For the best reading experience, check our web or app version and sign up for future editions here.

What’s cooking this week? 🥘

What's on the menu this week? Anthropic had the operational security week from hell (two leaks! including 512,000 lines of source code that competitors mirrored within hours), an NBER study told 89% of firms their AI isn't actually doing anything, and every major platform simultaneously decided your inbox/Slack/issue tracker needs an AI agent living inside it. Oh, and someone handed Claude $50K to play the stock market (because why not at this point).

Your Menu

  • 📰 Not Boring - Claude Code's 512K-line leak, platform wars (Gmail, Slack ship agents, Cloudflare challenges WordPress), the subprime technical debt crisis, 89% zero AI productivity impact, and OpenAI's expanding graveyard

  • ⌚️ Productivity Tapas - Cardboard (Cursor for video), Bagel AI (VoC to revenue), sitefire.ai (agentic web marketing), Goblin Tools (free task breaker)

  • 🍔 Blog Bites - Why AI makes your days harder (Cutler), pricing architecture that ships in hours (Huryn), and navigating org drama (Singhal)

  • 🎙️ Pod Shots - Morgan Housel on the psychology of money and why independence is the real wealth

Let's go 🚀

📰 Not boring

💰 Money Talks

$297B in one quarter. OpenAI worth more than all but 15 S&P 500 companies before turning a profit. The talent war ($400K retention bonuses!) tells you the real bottleneck isn't models - it's the people who build products around them. And Qodo's $70M bet on verification over generation might be the smartest contrarian play in the batch.

🤖 The Claude Code Chronicles

Anthropic's worst week for opsec became its best week for mindshare. The leak showed Claude Code is 99.96% scaffolding and 0.04% API call - the value in AI tooling is entirely in the harness, not the model.

🏗️ Platform Wars & Product Launches

Every major platform shipped AI this week - Gmail, Slack, Apple, Ring, Bluesky. Same playbook: embed intelligence, own the workflow, make switching costly. But the GitHub PR tips fiasco shows how quickly AI features destroy trust when the product judgement is wrong. And Cloudflare vs WordPress might be the quietest bombshell of the lot.

🧪 The AI Reckoning

  • The Subprime Technical Debt Crisis - AI-generated code seems manageable only if you believe models will keep improving fast enough to fix it. Same logic subprime lenders used about house prices. When the assumption breaks, the debt becomes unpayable

  • NBER study: 89% of firms report zero AI productivity impact - 6,000 executives across US, UK, Germany, Australia. 69% use AI, but executives average only 1.5 hours/week personally and predict a modest 1.4% productivity boost over 3 years

  • ARC-AGI-3 benchmark drops AI into a video game with no instructions. Gemini, Claude, ChatGPT, Grok all scored less than 1%. Most humans solve it easily. Memorisation isn't thinking

  • "I Quit. The Clankers Won." - a 15-year developer's resignation letter that became Hacker News' most-discussed post (390 pts, 453 comments): "organisations want your skills to atrophy as you instead use the AI machine to do the job faster"

*89% zero productivity impact. Models scoring below 1% on novel tasks. A developer manifesto with 453 comments mourning the craft. The subprime debt analogy is the sharpest framing yet: we're generating code fast under the assumption that future AI will clean it up.

🔧 Agentic Engineering

  • JustPaid created 7 AI agents grinding code 24/7 - 10 major features in a month. Plans to replace all employees with AI. Its latest human hire was trained almost entirely by AI agents

  • SaaStr: "We have 30 AI agents in production - here are the top 5 issues nobody talks about" - one agent autonomously offered free conference tickets (understood "discounts drive conversions" and ran with it). Management overhead: 20 agents x 20 min daily = 6.7 hours of babysitting. No agent-of-agents exists yet to manage the fleet

  • What is inference engineering? - 45-minute deep dive on quantisation, speculative decoding, prefix caching. Open models are 80% cheaper at scale. Plus 1-Bit Bonsai runs an 8B parameter model in 1GB of RAM (vs 16GB standard) - Apache 2.0, making serious AI possible on phones and edge devices

  • Harness designs domain-specific agent teams and auto-generates their skills. AgentField makes agents callable across your stack like any API. AI-Scientist-v2 generates hypotheses, runs experiments, and publishes peer-reviewed papers autonomously

JustPaid says "replace all employees." SaaStr says "we spend 6.7 hours a day babysitting ours." The gap between those two narratives is where every PM's agent strategy will actually land. Meanwhile, 1-bit LLMs in 1GB and open models at 80% less cost - the efficiency revolution is being driven by open source, not the labs charging you per token.

📱 The OpenAI Graveyard

The graveyard piece is essential reading for any PM working with AI labs as partners. If your roadmap depends on a partner's feature, ask: is this their core revenue driver, or a side quest? Side quests die.

🛠️ Tools & Launches

  • Autify Aximo navigates your app like a real user - describe what you want tested, get results. Adapts to UI changes across web, mobile, and desktop. AppDeploy ships ChatGPT/Claude apps to live URLs with zero infrastructure

  • Rovo connects docs, projects, code, decisions, and people into a single permissions-aware knowledge layer. Softr builds production-ready business apps from plain English

  • MiniMax M2.7 is the first model to participate in its own training, and open-sources their Office Skills library for Word, Excel, PDF, PPT. Microsoft's free AI agents course hits GitHub

  • Figma's next-gen data caching platform - a 23-minute deep dive worth bookmarking for any PM working at scale

Autify Aximo points the direction: AI that navigates real products, not just generates content. "Describe what you want tested, get results" will kill traditional QA scripting faster than most teams expect.

🌍 Big Picture & Security

Two supply chain attacks in one week. Your dependencies have dependencies, and state actors are actively poisoning the well. Every PM shipping software should be asking their engineering team: when did we last audit our dependency tree?

88% resolved. 22% stayed loyal. What went wrong?

That's the AI paradox hiding in your CX stack. Tickets close. Customers leave. And most teams don't see it coming because they're measuring the wrong things.

Efficiency metrics look great on paper. Handle time down. Containment rate up. But customer loyalty? That's a different story — and it's one your current dashboards probably aren't telling you.

Gladly's 2026 Customer Expectations Report surveyed thousands of real consumers to find out exactly where AI-powered service breaks trust, and what separates the platforms that drive retention from the ones that quietly erode it.

If you're architecting the CX stack, this is the data you need to build it right. Not just fast. Not just cheap. Built to last.

See the data

⌚ Productivity Tapas: Time-Saving Tools & Workflow Automation

  • Cardboard: "Cursor for video editing" - describe the edit you want in plain English and it executes. Browser-based, AI captions, beat-sync, highlight extraction. YC W26

  • Bagel AI: extracts Voice of Customer signals from Salesforce, Gong, Zendesk, and Jira - then connects feedback directly to revenue impact, churn risk, and active deals. No manual taxonomy

  • sitefire.ai: marketing for the "agentic web" - AI agents analyse what content drives LLM citations, write brand-aware articles, and optimise your visibility across ChatGPT, Claude, Gemini, and Perplexity. YC W26

  • Goblin Tools: free AI toolkit for breaking overwhelming tasks into manageable steps (Magic ToDo), rewriting tone (Formaliser), estimating time realistically (Estimator), and organising brain dumps (Compiler). No account needed

    Remember. Product Tapas subscribers get our complete toolkit - 550+ personally tailored, time-saving tools for PMs and founders. Your shortcut to efficiency and what's hot in product management 🔥

Check the link here to access.

🍔 Blog Bites - Essential Reads for Product Teams

Strategy: Institutionalised Overload - Why AI Is Making Your Days Harder, Not Easier

John Cutler examines why AI tools are reinforcing chaotic work systems rather than fixing them. Instead of reducing pressure, they help us manage more overload faster - turning busyness into a celebrated survival skill while genuine effectiveness suffers. Read the full article here.

💡 "Overload becomes so normalised and even celebrated that suggesting we do less, or process less, starts to sound almost heretical."

Why I picked this: This pairs directly with the HBR burnout research and the "I Quit. The Clankers Won." developer manifesto from this week's Reddit discoveries. If your team is "using AI to do more" but not doing better, Cutler explains why.

❝

Key Takeaways

• AI as coping mechanism, not cure : AI tools amplify existing demands by helping you juggle more inputs faster. The result is dependency on the technology to manage chaos rather than elimination of the chaos itself

•The normalisation trap : Success gets measured by your ability to handle increasing noise and demands - not by outcomes. This cultural shift makes it nearly impossible to argue for doing less

• Execution replaced by verification : What used to take 3 hours now takes 45 minutes to produce. But the interpretation, fact-checking, and decision-making around that output is more cognitively demanding - your days feel harder because they are

• The innovation illusion : Overload gets framed as essential for innovation ("we move fast!"), justifying chaotic conditions and stifling the organisational changes that would actually improve output quality

John Cutler, The Beautiful Mess

Strategy: Pricing Architecture - Four Pillars That Let Vercel Ship Price Changes in Hours, Not Quarters

Pawel Huryn and Fynn Glover (with input from pricing leaders at Superhuman, Stripe, Cursor, Vercel, and Supabase) make the case that pricing should live in config, not code. They lay out a four-pillar monetisation architecture that separates pricing logic from engineering, letting product teams experiment with pricing as fluidly as they do with features. Read the full article here.

💡 "Pricing should live in config, not code."

Why we picked this: With AI transforming cost structures across SaaS (token pricing, usage-based billing, hybrid models), the ability to iterate pricing as fast as features is becoming table stakes. Practical and immediately actionable.

❝

Key Takeaways:

• The four pillars : Unified product catalogue, decoupled entitlements, real-time metering, and a control plane. Together they create a system where pricing changes don't require engineering sprints

• Speed as competitive advantage : Companies with modern pricing architecture ship changes in hours. Those without it queue pricing behind feature work for quarters - and lose revenue every day they wait

• AI pricing is urgent : As AI transforms cost structures (tokens, compute, unpredictable usage), the old annual-review pricing cycle becomes actively dangerous. You need infrastructure that can adapt continuously

• Fear of churn costs more than churn : Analysis shows delaying price increases due to churn anxiety routinely costs more than the churn itself. The data consistently favours bold pricing moves on faster cycles

Pawel Huryn and Fynn Glover, The Product Compass

Leadership: How to Navigate Org Drama - The New Rules

Nikhyl Singhal (three-time founder, former CPO at Meta and Google) provides a tactical playbook for navigating reorgs, difficult managers, and organisational politics in today's flat, fast-moving companies. The key shift: strategic timing and delivery matter more than being right. Read the full article here.

💡 "The people who navigate org drama well today are strategic about when and how they act."

Why we picked this: Every PM has lived through at least one awful reorg. Singhal's advice is unusually tactical (not just "be resilient!") and comes from someone who's navigated the politics at Meta and Google. A palate cleanser from all the AI content this week.

❝

Key Takeaways:

•Assess before reacting : In flat orgs, evaluate whether your role keeps you relevant or risks you falling behind. Choosing to stay or go should factor both current situation and career trajectory

• Don't corner your boss : Before raising concerns with superiors, consider whether your feedback backs them into a position they can't solve. Timing and framing determine whether you get change or get sidelined

• Execute through chaos : Maintaining productivity during reorgs is your strongest protection. The people who keep shipping while others panic are the ones who survive restructuring

• Network before you need it : Build cross-functional relationships proactively - waiting until reorg rumours start means you're already too late. Genuine connections yield intelligence and allies

Nikhyl Singhal, The Skip

🎙 Pod Shots - Bitesized Podcast Summaries

Remember, we've built an ever-growing library of our top podcast summaries (120 or so). Whether you need a quick refresher, want to preview an episode, or need to get up to speed fast - we've got you covered.

Check it out here

🎙️ The Art of Spending Money

Morgan Housel's Psychology of Money sold millions of copies because it told a truth most finance advice ignores: our relationship with money is emotional, not rational.

In this conversation, he goes deeper on what we're actually seeking when we pursue wealth.

Why this matters for product people: Housel's framework isn't just personal finance - it's a lens on how we make career decisions, how we prioritise, and why so many PMs burn out chasing the wrong metric. If you've ever traded sleep for a launch, taken a role for the title, or wondered whether the next promotion will actually make you happier - this one's for you.

🎥 Watch the full episode here: Morgan Housel: Understand & Apply the Psychology of Money to Gain Greater Happiness
📆 Published: December 2024
🕒 Estimated Reading Time: 9 mins. Time saved: 111 mins! 🔥

Key insights from the full article:

  • 🎯 What we seek is freedom, not money — Independence - doing what you want, when you want - is what money actually buys. Most people confuse the tool with the goal.

  • ⚖️ Most people are at the extremes — Either saving too much (hoarding against anxiety) or spending too much (chasing dopamine). Balance is rare.

  • 💡 Money can buffer stress — The "money doesn't buy happiness" claim ignores that poverty creates stress. Financial security removes threats.

  • 🧠 Financial decisions are emotional — We justify with logic but decide with feeling. Understanding your emotional relationship with money matters more than spreadsheets.

  • 🔒 Freedom is independence — The ability to walk away, to say no, to wait - these options are what money actually provides.

  • 📊 Time is the real wealth — How you spend time matters more than how much money you have. Wealth is measured in optionality, not dollars.

  • 🚫 Don't follow someone else's playbook — Your psychology with money is unique. What works for others may harm you.

EXPLORE THE FULL GUIDE →

That’s a wrap.

As always, the journey doesn't end here!

Please share and let us know what you would like to see more or less of so we can continue to improve your Product Tapas. 🚀👋

Alastair 🍽️.

Reply

Avatar

or to participate

Keep Reading

© 2026 Product Tapas..
Report abusePrivacy policyTerms of use
beehiivPowered by beehiiv