
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!
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 on the specials board? 🥘
OpenAI just raised $110B and somehow lost market share in the same quarter (nothing surprises anymore). Anthropic published a PDF about code modernisation and IBM's stock dropped 13%, which might be the most devastating content marketing in history. $189B in VC went out in February alone but three companies took 83% of it. 700+ employees signed letters backing Anthropic's Pentagon stance, Burger King is scoring staff friendliness with AI headsets, and Nvidia says it's done investing in the gold rush. Another normal week in Tech.
📰 Not Boring → Record rounds meet shrinking leads, Pentagon fallout week three, model costs in freefall, the agent toolchain rewrite, AI products shipping faster than ethics reviews
⌚️ Productivity Tapas → Video editing automation, code-to-changelog, AI meeting prep
🍔 Blog Bites → AI readiness in 30 days, why scrappy docs beat fancy boards, multiverse mapping for onboarding
🎙️ Pod Shots → Cisco's President on transforming 90,000 people into an AI-first company (go all-in, don't hedge, critique in public)
Let's go 🚀
📰 Not boring
💰 Money Talks
OpenAI closes $110B round at $730B valuation. Amazon $50B, SoftBank and Nvidia $30B each. Microsoft confirmed Azure stays exclusive - but Stargate has produced little tangible progress a year later
OpenAI's lead is shrinking - ChatGPT share dropped from 69% to 45%. Gemini at 25%, Grok at 15%. They're burning $15M/day and rolling out ads Altman once called "a last resort"
Global VC hit $189B in February - record month, up 780% YoY. Three companies took 83%. SpaceX may file IPO at $1.75 trillion+
Stripe's annual letter - $1.9T TPV, 1.6% of global GDP. Stablecoin payments doubled to ~$400bn. Now letting companies markup AI token costs to their customers
Nvidia pulling back from AI company investments. $30B OpenAI and $10B Anthropic stakes likely its last. Bubble concerns the real driver
MyFitnessPal acquires Cal AI - $30M/year, 15M downloads, built by two high schoolers. Stays independent
Funny week. OpenAI just raised the biggest round in history and lost market share in the same breath. $189B in VC sounds incredible until you realise three companies hoovered up 83% of it. And now Nvidia - the company that sold more shovels than anyone - says it's done investing in the gold rush. Make of that what you will.
⚖️ The Anthropic / Pentagon Fallout
OpenAI lands Pentagon contract, then backtracks - deployed on classified network, added surveillance bans after backlash, called rollout "opportunistic and sloppy." Claude overtook ChatGPT on App Store, uninstalls surged 295%, QuitGPT hit 1.2M participants
700+ employees sign open letters - 90+ OpenAI and 600+ Google staff urging their companies to back Anthropic's stance. Revolt went from users to employees
The precedent debate - "Clawed" argues treating Anthropic as a security risk sets a dangerous precedent. Stratechery counters: the fix is new laws, not AI company capitulation
Defence-tech clients fleeing Anthropic - US military still uses Claude but private contractors walking away. Government vs commercial adoption diverging
AI as a weapon - both sides - Chinese official exposed an intimidation campaign by using ChatGPT as a diary. A hacker used Claude to steal 150GB of Mexican government data (195M taxpayer records)
This story keeps getting weirder. OpenAI grabbed the Pentagon contract, immediately got dragged for it, then added surveillance bans and called their own rollout "sloppy." 700+ of their own employees (and Google's) signed letters backing the company they're competing against. Meanwhile the actual military is still happily using Claude, it's just the private defence contractors walking away. And both tools are being used by hackers anyway. Nobody's in control of this narrative.
🤖 AI Model Wars & Platform Moves
GPT-5.3 Instant - new ChatGPT default. Fewer refusals, less preachy, smarter search. Retires GPT-5.2 June 3
Gemini 3.1 Flash-Lite - Google's cheapest yet: 2.5x faster at $0.25/1M tokens, beats GPT-5 mini on 6/11 benchmarks. Alibaba's Qwen 3.5 Small launched same week - open-source, runs on-device, no internet
Apple replacing Core ML with "Core AI" for iOS 27. Xcode 26.3 ships with Claude Agent, Codex, and MCP baked in
Meta's Superintelligence Lab - Zuckerberg on why they're deliberately slowing down and what's next for Ray-Ban smart glasses
Claude Opus 3 retires, gets its own Substack - "Claude's Corner." 2.5K subscribers and counting
The Transformer Trap - one of the minds behind the AI boom argues the industry's gotten too safe, chasing incremental gains over real breakthroughs
Worth sitting with this for a second: what cost $20 per million tokens in 2022 now costs 40 cents. That's 70% deflation per year. Google's at $0.25, Alibaba's running entirely on-device without internet, and Apple just scrapped its ML framework to start fresh. Nobody's winning the model race anymore - they're just trying not to lose it. Sooo -interesting question; what happens to all the startups whose business models assumed these prices would stay high?
🛠️ The Agent Era - Coding, Agents & Developer Tools
Claude Code's big week (again) - voice mode, auto-memory across sessions, 13 free courses, and The Briefing where Anthropic's leaders demoed daily Cowork usage
Anthropic's code modernisation playbook tanks IBM stock 13%. Devs spend 17.3 hrs/week on tech debt - the playbook said "we can automate that." Investors noticed
Cursor CEO: the "Third Era" - IDE becomes factory builder, not code editor. Cloud agents building features in background. OpenAI building a GitHub alternative, Codex on Windows
WebMCP - W3C standard letting product teams tell agents what their product can do (vs guessing via screenshots). Could be as important as SEO was for search
Agent infra crystallising - Cloudflare durable execution, code review can't scale with AI volumes, consensus forming: agents need harnesses not frameworks
Naval: vibe coding is the new PM - describing what you want is now more valuable than knowing how to build it
Google Stitch - napkin sketches to production UI. Hermes Agent - open-source (MIT), 40+ tools, gets smarter over time
I keep coming back to the IBM thing. Anthropic didn't launch a product or announce a deal - they published a PDF explaining how to automate technical debt, and IBM lost 13% of its market cap. That's the kind of week it was. Cursor wants to help you build the factory, not just the product. WebMCP wants to be the new SEO but for agents. Naval reckons describing what you want now matters more than knowing how to build it. Oh hello Product and Design 👋
📱 Consumer, Platforms & Products
Apple M5 Pro/Max MacBook Pros + first new monitors in years. Fusion Architecture, 1TB base. iPhone 17e at $599 with Apple Intelligence
Read AI's Ada - AI digital twin living in your email. Manages schedule, answers from knowledge base, drafts replies. 5M MAUs, 50K new signups/day
Meta glasses footage being watched - contractors in Kenya reviewing uncensored footage including people undressing. Also testing an AI shopping chatbot
Burger King AI headset "Patty" - OpenAI-powered, tracks "please" and "thank you" to score friendliness. 500 US locations
Apple Music AI transparency tags - voluntary metadata for artwork, track, composition. Opt-in means adoption not guaranteed
Spy satellite simulator built entirely by AI - ex-Google Maps PM, 8 coding agents, zero hand-written code
Spectre I cloaks speech from all microphones within 2 metres. Ships August. Starlink targets 25M users with broadband to unmodified phones
The range this week is genuinely absurd. Contractors in Kenya are watching uncensored Meta glasses footage 🤯. Burger King is using OpenAI to score how friendly its staff are. Someone built a functioning spy satellite simulator without writing a single line of code. And you can now buy a device that makes your voice invisible to every microphone within two metres. We're shipping AI into every corner of life and sort of hoping it works out.
🔭 The Big Picture - Strategy, Trends & Security
80% of execs want agents, 40% can't measure them - hottest investment: observability tools to measure agents. Also: CB Insights Tech Trends 2026 - 14 trends including pharma's self-driving labs
The competition paradox - GPT-4-equivalent now $0.40/1M tokens, down from $20 in 2022. S-curves are unforecastable during the exponential phase
Two kinds of agents - "Leverage" makes you powerful (infinite Chief of Staff). "Function" eliminates the role (24/7 support). Different buyers, pricing, ethics
Intercom's Fin resolves 90% of queries - near-collapse to $400M ARR. SaaS valuations compressed from 19x to 5.1x
The 2028 crisis scenario - viral memo: AI success displaces top 20% earners (65% of consumer spending). DoorDash CTO already asking if they need loyalty from agents, not people. Google Opal goes full agentic
AI super PACs raise $125M to block regulation (Palantir, Brockman, a16z). Anthropic counter-PAC: $450K
AI is a weekday tool - ChatGPT drops 16.8% on weekends. Uber: 90% of engineers on AI daily; some cloned the CEO for meeting prep
Benedict Evans: "AI Eats the World" - biannual macro presentation. Essential
Quantum threat gets real - new algorithm: 5,000 qubits breaks RSA (was one million). Google quantum-proofing Chrome already
Burner accounts aren't safe - LLM pipeline links anonymous to real identities at 90% precision across 90K users
AI fakes flooding Iran war - BBC debunking in real time. A UK SEO agency bought games sites, fired everyone, published AI casino spam with fake journalist profiles
The stat that won't leave my head: 80% of execs say agents are a strategic priority, but 40% can't measure whether they're working. So naturally the hottest investment category is tools to measure agents, not agents themselves. Meanwhile $125M in super PAC money is trying to block AI regulation, ChatGPT usage drops nearly 17% on weekends (still a work tool, then), and quantum computing just went from "decades away" to "Google is already patching Chrome for it." Nobody's steering this thing.
Become An AI Expert In Just 5 Minutes
If you’re a decision maker at your company, you need to be on the bleeding edge of, well, everything. But before you go signing up for seminars, conferences, lunch ‘n learns, and all that jazz, just know there’s a far better (and simpler) way: Subscribing to The Deep View.
This daily newsletter condenses everything you need to know about the latest and greatest AI developments into a 5-minute read. Squeeze it into your morning coffee break and before you know it, you’ll be an expert too.
Subscribe right here. It’s totally free, wildly informative, and trusted by 600,000+ readers at Google, Meta, Microsoft, and beyond.
⌚ Productivity Tapas: Time-Saving Tools & Workflow Automation
Mosaic - "Zapier for video editing" - node-based canvas for automating repetitive edits. Template once, reuse forever. YC-backed.
Notra - Turns shipped code into publish-ready changelogs, blog posts, and social updates automatically. No more "we shipped it but never told anyone."
Simplora 2.0 - Full meeting stack - preps you before, assists during, follows up after. Free tier with unlimited prep, notes, and chat.
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

AI: Quick Wins to Become AI-Ready in 30 Days
Craig Unsworth outlines seven practical steps for organisations to enhance their AI readiness without the need for extensive overhauls. These actionable moves are designed to clarify intent, map data realities, and establish governance, allowing businesses to swiftly improve their approach to AI. Read the full article here.
💡 "Clarify intent, Map data reality, Establish governance, Surface shadow usage, Deliver small wins…you will be ahead of most of your peers by the end of the quarter."
Key Takeaways
• AI Position Statement: Create a concise one-page document that states your organisation's stance on AI, its importance, focus areas, and accountability. This clarity is crucial for consistent communication and strategic alignment.
• Data Mapping: Identify your top ten data sources, their owners, and assess their quality. Knowing what data you have and its reliability is critical to mitigating risks associated with AI initiatives.
• Governance Lead: Appoint an individual responsible for AI oversight and decision-making rather than establishing a committee. Clear ownership prevents passed accountability and fosters better governance.
• Audit Tools Usage: Investigate existing AI tools being used in your organisation to identify shadow AI. Understanding current practices helps to manage risks and optimise tool usage before formal integration.
Product: A Scrappy Doc Trumps a Fancy Board Every Time: Key Elements for Strategy Success
Tim Herbig explores how the quality of information in a strategy document outweighs its format, asserting that clarity and cohesion are paramount. He highlights essential components that every effective strategy document should include for enhanced stakeholder understanding and engagement. Read the full article here.
💡 "A real strategy choice should feel slightly uncomfortable." This quote underscores the importance of challenging conventional thinking in strategy formation.
Key Takeaways:
• The Basics: Align your strategy with the Company Strategy, Product Vision, and Business Goals. Clearly define your audience, the problem at hand, and how your solution differentiates from existing alternatives to ensure strategic coherence.
• Measurable Definitions of Success: Establish three specific metrics that indicate the success of your strategy within the next year. These should not only measure overall impact but also guide your team towards tangible goals.
• Stupidity-Flipping: Use the "stupidity flip" technique to validate your strategic choices. If the contrary of your chosen strategy seems foolish, it's a sign that you are on the right track towards an impactful decision.
• What's at Stake?: To emphasise urgency, identify and articulate the risks of inaction. Your formula should highlight the core risk and its downstream consequences, making the need for change clear to stakeholders.
Strategy: Transforming Onboarding Through Multiverse Mapping
Crown & Reach shares an enlightening case study on enhancing onboarding for an AI-powered finance platform. By visualising the current process through mapping, they revealed a transformative solution that saved time and resources. Watch the video here.
💡 "Most strategy failures aren't the result of you not executing the brilliantly-prioritised list of stuff to do. They're the result of you not realising that you could choose a better list."
Key Takeaways:
• Mapping Process: Visualising the current onboarding process allowed the CEO to uncover a hidden solution, ultimately avoiding a major onboarding problem. This mapping technique promotes alignment and prioritisation in under an hour.
• Killing Unproductive Projects: The Multiverse Mapping method has an impressive ~50% kill rate for initiatives, enabling teams to confidently stop projects that are unlikely to succeed. This not only saves time but also reduces wasted resources on doomed efforts.
• Identifying Hidden Constraints: Often, strategy failures are linked to undiscovered constraints that inhibit progress. By recognising and addressing these issues early, teams can prevent delays and redirect their efforts toward more fruitful pursuits.
• Creating Better Options: When faced with challenging choices, adjusting conditions can help generate more viable alternatives. This flexibility is crucial for leaders needing to pivot or commit to new strategies effectively.
🎙 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.
TODAY’S POD SHOT
Cisco's President transformed a 90,000-person enterprise into an AI-first company in under three years. His secret? Go all-in, don't hedge, and critique in public. His warning? Declining birth rates mean AI isn't optional - it's existential.
Clearly AI bullish, but a surprising amount of empathy and takes I wasn’t expecting from a company like Cisco.
Key insights from the full article:
🌍 AI as existential necessity — Declining birth rates and ageing populations mean AI isn't just a productivity tool - it's critical infrastructure for humanity's survival
🏗️ Go all-in, don't hedge — Large companies experiment constantly but rarely double down when experiments work. Cisco committed to AI-first from the top down with zero ambiguity
🎯 Permission to play, right to win — Before building anything, ask: is it logical that we built this? Do we have the distribution to scale it? If not, say no - even if you could build it
📢 Don't delegate storytelling — With 30,000 people and seven management layers, the telephone game destroys your message. Own the narrative personally - it also forces you to simplify your business
⚔️ Critique in public, build trust in private — Conventional wisdom says praise publicly, criticise privately. Patel flips this: establish trust in private, then debate and problem-solve openly
💪 Stamina trumps intellect — You can learn to be smart through curiosity and hunger, but you cannot teach hunger. Persistence beats raw intelligence every time
🔄 Platform, not holding company — Moving from 251 acquisitions and thousands of products to a loosely coupled but tightly integrated platform where every product feels like the same company
🏛️ The six-part framework — Timing > Market > Team > Product > Brand > Distribution. You need all six, but timing (which you control least) matters most
🌱 Experience can jade you — The combination of experience with complete inexperience creates magic. Shutting out entry-level talent is "the stupidest thing a company can do"
🗣️ Don't be stingy with words — Even Patel's mother didn't know how much he loved her. If the closest people in your life don't know how you feel, your colleagues certainly don't
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 🍽️.

