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!

Welcome to this week’s 🌮 Product Tapas!

New here? We're your shortcut to staying sharp. Essential stories, practical tools, real insights. If you’ve been forwarded this or just stumbled upon it, you’re in for a treat. For the best reading experience, check out the web version and sign up for future editions here.

What’s sizzling this week? 🥘 

Agent hype intensifies 🫨 - Notion's running 300 agents for 1,000 employees and Google's computer-use models can organise sticky notes (finally, AI for Post-its). AI infrastructure costs keep inflating with $500B a year needed just to keep up (totally sustainable), and Deloitte's forced to refund for AI hallucinations in client work (totally unrelated to last week's full colleague roll-out news, obvs). Meanwhile your face is officially an API now (open-source yourself).

• 📰 Not Boring → Agent orchestration wars, Google's AI feature flood, infrastructure bubble math

• ⌚️ Productivity Tapas → AI agent studios, visual remix tools, product roadmap intelligence

• 🍔 Blog Bites → WhatsApp's phone number genius, the power of strategic silence, why traditional companies struggle with product transformation

• 🎙️ Pod Shots → Google's $4 trillion mistake (how the inventor of modern AI got caught napping)

Let's go 🚀

📰 Not boring

Agents Everywhere

• For those that want to go DEEP - the guy who shipped 3 AI agents at Google I/O 2025 just open-sourced his 424-page design playbook. Production-ready code for 21 patterns across the entire stack. Free via Google Drive. Here

• Google launches Gemini Enterprise to help companies automate tasks and deploy AI agents across all departments, code-free. It integrates with major platforms like Salesforce and Microsoft 365, using internal data and Google AI to generate content and extract insights, as seen with early adopters like Figma and Mercedes

• More on Google's new computer use model—here's a demo of it updating CRM data and here organising sticky notes after a meeting

• Amazon reboots its AI Agent for business, taking on ChatGPT, Copilot

• Slack is rolling out new AI features, powered by OpenAI, Anthropic, and others, to help users draft plans, summarise reports, and prioritise tasks directly within the platform. This includes an enhanced Slackbot and a ChatGPT app, allowing teams to leverage AI assistants and integrate with tools like Google Drive and Salesforce

• Notion has over 300 custom agents working alongside its 1,000 employees, supposedly completely changing the way they work

• Square launches AI voice ordering and an integrated Bitcoin solution for merchants

• Copilot now works with Google and MS Office

• AI browser Dia launches for free to Mac users

We're deep in the "agents everywhere" phase. Google's shipping computer-use models that update CRMs and organise sticky notes; Notion's running 300 agents for 1,000 employees. The centre of gravity is shifting from "one assistant to rule them all" to "agents everywhere you already work." Google and Amazon are selling orchestration at the top of the stack; Slack, Notion, and Square are pushing task-level agents at the edge. Whoever owns the agent policy layer will quietly own the enterprise.

Google Floods the Zone

• Google is adding AI across its lineup: Search gets "hide sponsored results" and AI Overviews; Gmail's Gemini suggests meeting times from context; Workspace adds Nano Banana for image edits; NotebookLM supports micro‑videos; and Meet introduces persistent virtual makeup filters

• Google's virtual try-on shopping tool expands to more countries, now lets you try on shoes

Classic Google strategy: ship a thousand small affordances until the aggregate feels like a platform shift. Each feature is small—meeting scheduling, image editing, makeup filters—but together they normalise AI having access to everything you do in Google's ecosystem.

The Infrastructure Crunch

• AI compute demand is now growing at over 2 TIMES the rate of Moore's Law, creating a massive shortage. Just to meet current demand, $500 billion must be invested in data centres PER YEAR until 2030

• Retool survey suggests 65% of internal products are now built by non-engineers (nb may be skewed by their low-code user base)

• n8n secures $180 Million Series C funding, hits $2.5 billion valuation to lead AI-powered workflow automation

The numbers are wild: $500 billion a year needed just to keep up with demand, workflow automation tools hitting $2.5 billion valuations, supposedly 65% of internal products now built by non-engineers. We're clearly in a bubble. Whether it's got years left to run or months, nobody knows—but the money keeps pouring in and the bets keep getting bigger.

Odds and Ends

• Ring launches Search Party—feature that goes through footage from nearby Ring outdoor cameras when a dog goes missing (on by default though…)

• Deloitte was caught using AI in $290,000 report; now forced into a partial refund after a researcher flagged hallucinations. Unrelated to last week's story of Deloitte rolling out AI to 470k employees of course. Oops

• "Your face isn't just a platform anymore. It's an API. The only question is whether you'll open-source it". This is a great piece on content, reach and the direction of travel enabled by AI video

• 3 principles for building great AI products (from the Claude Code team): Build for yourself, creating tools you personally need; design for a simple start with powerful, extensible capabilities; and let latent demand guide your roadmap by turning user "hacks" into new features

• 50 creative ways non-tech users are getting the best out of Claude Code

Ring's Search Party defaults everyone into neighbourhood surveillance—helpful for lost dogs, less so for privacy. Deloitte's AI hallucination refund is a nice bookend to last week's 470k employee rollout. And "your face as an API" captures where AI video is heading: identity becomes programmable infrastructure.

The Gold standard for AI news

AI keeps coming up at work, but you still don't get it?

That's exactly why 1M+ professionals working at Google, Meta, and OpenAI read Superhuman AI daily.

Here's what you get:

• Daily AI news that matters for your career - Filtered from 1000s of sources so you know what affects your industry.

• Step-by-step tutorials you can use immediately - Real prompts and workflows that solve actual business problems.

• New AI tools tested and reviewed - We try everything to deliver tools that drive real results.

• All in just 3 minutes a day

Join 1M+ pros

⌚️ Productivity Tapas: Time-Saving Tools & GPTs

• Boomi Agentstudio: securely design govern and orchestrate AI agents at scale

• Masonry: Create, remix, and refine visuals with every major AI model in one elegant workspace

• Squad: Your AI product manager (analyse your data, surface insights,
  and build your product roadmap)

• Convert any PDF into a fillable form with AI magic

• VibePM Gamma tips: check out Dave Killeen’s latest video on how to get the best out of Gamma. Nestled in the middle of it is the GREAT tip to use the Anthropic Console to enhance your prompts. It’s is literally THE BEST way I know to enhance your prompts and it costs $5 for about 6 months of credits

  Remember. Product Tapas subscribers get our complete toolkit - 460+ 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: How WhatsApp Cracked the Cold Start Problem with Phone Number Genius

Tom from Strategy Breakdowns explores WhatsApp's revolutionary signup strategy that eliminated traditional onboarding friction by using phone numbers instead of email addresses. The approach transformed messaging from "yet another account to manage" into a natural upgrade of your phone's core function. Read the full article here.

💡 "We created a unique thing: we built the messenger connected to contacts from a phone book. ICQ, for example, gave you a 7-digits number you had to memorise, write on a paper and tell your friends. It was not very user-friendly, to put it mildly."

Communication: Why Silence is Your Secret Weapon

Jaryd Hermann explores the counterintuitive power of restraint in our oversharing culture. He argues that saying less—both about our plans and in general conversation—can dramatically increase our focus, follow-through, and influence. Read the full article here.

💡 "Saying you will do something is not the same as doing something. And the people who end up getting the most done and have the most follow through tend not to overshare their plans too soon."

This insight challenges our social media-driven impulse to announce every intention, revealing how premature sharing can actually sabotage our goals. It’s an interesting take - especially as I’ve seen people gain drive and achieve more from publicly committing to do things. I guess balance is key 🤔 

Strategy: Why Traditional Companies Struggle with Product-Centric Transformation

John Cutler explores the unique challenges non-digital companies face when adopting product-centric ways of working. He argues that simply copying digital-first models isn't enough—complex enterprises need a more nuanced, ecosystem-based approach that embraces multiple operating models simultaneously. Read the full article here.

💡 "The real opportunity lies in embracing a more networked, ecosystem-based approach fully. You have to accept that multiple motions will operate at once."

This insight challenges the common "one-size-fits-all" approach to product transformation, recognising that complex organisations need sophisticated operating models rather than simplified frameworks.

🎙️ Pod Shots - Bitesized Podcast Summaries

Remember, we've built an ever-growing library of our top podcast summaries. 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

🤖 Google's $4 Trillion Mistake? How the Inventor of Modern AI Got Caught Napping

Link below to the full summary - but here’s what’s in store:

"Google invented the Transformer. OpenAI made it a household name."

Google assembled the greatest AI talent in history, built the architecture powering ChatGPT, and deployed language models at scale in 2001—two decades before the AI hype cycle. Then ChatGPT launched, and they were caught flat-footed. How did the company that created modern AI get caught napping?

Key insights from the full article:

• 🤖 Google's $4 trillion mistake — inventing the future doesn't guarantee winning it

• 🧬 Compression = intelligence — Google's 2001 insight that predicted modern LLMs

• ⚡ Jeff Dean's superpower — reducing translation time from 12 hours to 100ms

• 🧠 The talent exodus — how OpenAI poached Google's best researchers in 2015

• 💎 2-3 million TPUs — Google runs an almost-Nvidia-scale chip operation internally

• 🔮 Eight researchers, one paper — the Transformer that changed everything

• 🚪 Strategic catastrophe — publishing the Transformer openly armed every competitor

• 🚨 Code red moment — ChatGPT was Google's "iPhone moment"

• 📊 The innovator's dilemma — protect $140B in search revenue or cannibalise it for AI?

👉 Read the full breakdown — sent separately, check your inbox

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 🍽️.