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.
For the best reading experience, check our web or app version and sign up for future editions here.
What’s cooking this week? 🥘
Apple's paying Google a billion for Siri whilst Google pays them twenty billion for search (circular economy goals), AI shopping assistants are getting sued faster than they can launch (Amazon's not threatened, obviously), and Brussels is quietly dismantling GDPR because falling behind finally matters more than protecting citizens. Meanwhile three startups just raised $200m betting that making complex things simple beats making simple things complex.
📰 Not Boring → AI model musical chairs, shopping wars contradictions, startup funding rounds
⌚️ Productivity Tapas → Free frontier models, voice-first wearables, automated IT support
🍔 Blog Bites → Lovable's pre-paywall hooks, OKR theatre warning signs, eBay's trust blueprint
🎙️ Pod Shots → Which AI model actually wins at your job (spoiler: it depends)
Let's go 🚀
📰 Not boring
Do the AI Model Shuffle 💃
Apple nears $1 billion-a-year deal to use Google AI for Siri (remember Google pays Apple ~$20bn p/a so it can be the default search engine on iPhones)
Microsoft lays out ambitious AI vision, free from OpenAI; revised deal allows Microsoft to pursue AGI/superintelligence after previously being contractually blocked, whilst maintaining 27% stake and IP rights through 2032
Gemini Deep Research can now connect to your Gmail, Docs, Drive and even Chat. Although Claude Code could do all of those apart from chat anyway…
Apple's paying Google a billion to power Siri whilst Google pays Apple twenty billion to be the default search engine on iPhones. Microsoft's contractually escaping OpenAI's grip to build AGI solo. Google's integrating Gemini everywhere - Gmail, Docs, Drive, Chat. The old "build vs buy" question has morphed into "whose model do we license this quarter?" Nobody wants to be caught holding yesterday's LLM.
AI Tools Leave the Lab
The demo-to-deployment gap is finally closing. Windsurf is making codebases legible through AI-annotated maps (product managers everywhere just locked in). Meanwhile Amazon's legal team is busier than their checkout-free stores, proving the real battle isn't technical capability - it's whose business model survives contact with AI shopping assistants.
Tech Giants Play Musical Chairs
Google settlement with Epic caps Play Store fees, boosts other Android app stores
WhatsApp finally lets you message people on other apps
Netflix is reportedly going big on video podcasts for 2026
Snap and Perplexity partner to bring conversational AI search to Snapchat (at a cost of $400M to Perplexity)
Platform strategies are splintering in opposite directions. Google's conceding ground to Epic on app store fees, WhatsApp's opening to cross-platform messaging (thank you Brussels), Netflix is chasing YouTube's billion monthly podcast viewers, and Perplexity's paying $400m to reach Gen Z through Snapchat. Nobody's following the same playbook anymore because there isn't one - some walls are crumbling, others are being rebuilt, and Perplexity just paid $400m to prove distribution still matters more than technology.
AI Shopping Wars Heat Up
Pinterest launches "Pinterest Assistant," their AI-powered tool to help shopping and discovery. Their visual-first AI outperforms off-the-shelf models by 30% on shopping recommendations
McKinsey warns Agentic commerce will kill advertising revenue by 2030 as AI agents bypass retailers
Whereas, IAB consumer study shows the opposite happening now: AI triples traffic to retail sites because shoppers don't trust it (only 46% do) and need to validate everything 🤷
McKinsey predicts doom for retail advertising whilst IAB's data shows AI is actually driving more traffic to stores because nobody trusts it yet. Pinterest is betting on visual search winning the "I'll know it when I see it" challenge that defeats text-based AI. One's predicting disruption, another's measuring expansion, and the third's building for a different shopping behaviour entirely. Everyone's right about their own timeline.
Startup Funding Rounds
Wabi raises $20m to create the YouTube for apps 👀
Gamma hits $2.1B valuation after recently raising $68M in fresh funding, and hitting $100M annual recurring revenue. Here's a recent prompt deck to help spark creativity and how to use it
Scribe hits $1.3B valuation as it moves to show where AI will actually pay off. After helping thousands of enterprises document how work actually happens, they're now identifying what's worth automating
Wabi's democratising app creation (YouTube but make it software), Gamma's worth $2.1B for fixing PowerPoint's 35-year blank slide problem, and Scribe's solving "what should we automate first?" The "AI for X" formula still prints money when X is genuinely painful - whether that's slide layouts, workflow documentation, or app distribution. Combined they've raised nearly $200m betting that making complex things simple beats making simple things complex.
Everything Else
McKinsey has released its State of AI in 2025: "Agents, Innovation, and Transformation". Key findings: AI's wins are mainly qualitative not quantitative; a small group is cracking the code; they redesign company-wide workflows; the winning playbook has 3 best practices
Here's an interesting paper on the impact of how people perceive ads that use GenAI images. TL;DR: AI-generated ads outperform human-created ones by 19% in click-through rates, but disclosing AI involvement tanks effectiveness by 31.5%
Google Labs releases "Learn Your Way," a new feature that rewrites books based on students' interests—leading to 11ppt uplift on tests
Brussels ready to cut back on some EU privacy rules to stop Europe from "falling behind" in AI boom
Here's an "MCP Security Best Practises Cheat Sheet" for you
Tesla shareholders approve Elon Musk's $1 trillion pay package (casual)
Bringing last week's Gemini <> Google Maps update into reality: you can now cross-reference places with Street View pics. So instead of "Turn left in 100m," you get "turn left after the post box"
Google Pixel update adds battery-saving maps mode, AI photo remixing, and smarter notifications
McKinsey's confirming my own bias: AI's value is mostly qualitative, concentrated in companies who've redesigned entire workflows rather than sprinkling AI on existing processes. The disclosure paradox is real—AI-generated ads work brilliantly until you admit they're AI-generated. Brussels is temporarily dismantling GDPR because "falling behind" finally outweighs "protecting citizens." And Google Maps now gives directions like your mum, which is genuinely more useful than coordinates.
And Finally...
AI turns brain scans into full sentences. We're all doomed
Japanese researchers are decoding visual thoughts from MRI scans with 50% accuracy - matching generated descriptions to the right video half the time. It's either revolutionary accessibility technology for people who've lost speech, or the privacy nightmare to end all privacy nightmares. Probably both. Either way, "I was just thinking about buying those shoes" just became a viable legal defence.
Voice AI Goes Mainstream in 2025
Human-like voice agents are moving from pilot to production. In Deepgram’s 2025 State of Voice AI Report, created with Opus Research, we surveyed 400 senior leaders across North America - many from $100M+ enterprises - to map what’s real and what’s next.
The data is clear:
97% already use voice technology; 84% plan to increase budgets this year.
80% still rely on traditional voice agents.
Only 21% are very satisfied.
Customer service tops the list of near-term wins, from task automation to order taking.
See where you stand against your peers, learn what separates leaders from laggards, and get practical guidance for deploying human-like agents in 2025.
⌚ Productivity Tapas: Time-Saving Tools & Workflow Automation
CTO.new: Code with the latest frontier models from Anthropic, OpenAI and more. FOR FREE
Sandbar: Wearable / voice first ring - record voice memos, create notes and generally record your thinking. Productivity / wearables will be massive.
Console: Automated IT support; auto-resolve repetitive IT requests, using natural language, not code.
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
Product: How Lovable Hooks Users Before They Hit the Paywall
Peter from Built for Mars explores the addictive user experience design of Lovable (AI-powered app building platform if you somehow missed it…!). Despite spending 60% of his time pointing out bugs during a 100-minute session, he found the experience surprisingly enjoyable and compelling. Read the full article here.
💡 "I spent 100 minutes building something with Lovable, and 60% of my time was spent just pointing out bugs. Which sounds like product hell. So why did I enjoy it so much?"
Key Takeaways
• First Impression Management: Lovable carefully orchestrates the initial user experience to create immediate engagement and investment before users encounter limitations or technical issues
• Friction vs. Engagement Balance: The platform manages to maintain user interest despite significant bugs and technical problems through strategic UX design and expectation setting
• Pre-Paywall Psychology: The tool employs specific tactics to build user commitment and perceived value before presenting upgrade options, making conversion feel natural rather than forced
• AI Tool Positioning: Lovable demonstrates how AI-powered development tools can create compelling user experiences even when the underlying technology isn't perfect
• Growth Hacking Methodology: The case study reveals systematic approaches to user acquisition and retention that can be applied across different product categories
Product Management: When OKR Planning Becomes Counterproductive Theatre
Tim Herbig explores the critical moment when OKR planning shifts from strategic clarity to wasteful process obsession. He introduces the "10% rule" as a practical benchmark for keeping planning focused and actionable. Read the full article here.
💡 "The moment you're tweaking wording instead of committing to a strategic goal, you've crossed from Real Progress into Alibi Progress."
Key Takeaways:
• The 10% Rule: OKR planning should consume roughly 10% of your cycle time—about 52 team hours quarterly. Warning signs you've exceeded this include weeks in alignment workshops, endless revisions, and debating word choices rather than strategic direction.
• Good Enough Test: OKRs pass when team members can explain the intent, they help say no to irrelevant work, and each Key Result traces back to a strategic choice. Perfect OKRs aren't the goal—directionally useful ones are.
• The Right Metrics: Focus on answering "X months from now, which three metrics would tell us this strategy choice has worked?" rather than tracking everything possible.
• Future-Proofing Exercise: Predict check-ins 4 weeks ahead by asking "With whom will you review these OKRs and what discussions will emerge from changed values?" This premortem helps test whether OKRs will drive actual decision-making.
• Root Cause Reality: Excessive OKR planning signals deeper organisational issues—uncertainty about strategic priorities or fear of committing to measurable outcomes—not problems with the framework itself.
• The Real Solution: It's not better templates or sophisticated tooling. It's having courage to say "these are directionally good enough, let's start."
• OKR Theatre Warning: When the process becomes the product, you're masking strategic uncertainty with procedural complexity.
Strategy: How eBay Cracked the Code on Digital Trust Between Strangers
Strategy Breakdowns explores how eBay transformed peer-to-peer commerce from impossible to scalable by manufacturing trust in a zero-trust environment. The analysis reveals how eBay pioneered the trust mechanisms that every modern marketplace platform still follows today. Read the full article here.
💡 "Create accountability mechanisms that align incentives and make honesty the only rational strategy."
Key Takeaways:
• Quantified Trust: Two-way feedback systems made trust measurable. Gamified star badges (up to silver shooting stars at 1M points) incentivised positive behaviour.
• Reputation Lock-In: Sellers with thousands of ratings couldn't move to competitors without losing their advantage—creating powerful switching costs.
• Layered Protection: PayPal brought fraud detection and buyer protection up to $500. Dispute resolution evolved through SquareTrade to integrated Resolution Centre.
• Global Infrastructure: Cross-border currency conversions and refund protections enabled worldwide peer-to-peer commerce that traditional banking couldn't mediate.
• Network Effects Moat: More trust data attracted more users, generating even more trust data—creating unbreachable competitive advantage.
• Market Transformation: Before eBay, peer-to-peer meant mailing cash to strangers with no recourse. eBay proved strangers could transact safely online.
• The Blueprint: Every modern two-sided marketplace from Airbnb to Uber follows eBay's playbook: quantified reputation, protection mechanisms, dispute resolution.
• Amazon's Counter-Strategy: Amazon dethroned eBay by owning the entire experience rather than leaving it to individual sellers—exposing marketplace model limits.
🎙 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
💡 Stop Wasting Time Comparing AI Models - Here's Which One Actually Wins at Your Job
Second one from Peter Yang in two weeks - sorry, (not sorry)! As I’m going deeper on AI tooling for a few weeks it helps take people on the journey…
"Don't use one AI for everything—you're leaving productivity on the table."
This practical guide cuts through the endless "GPT-5 vs Claude vs Gemini" debate and reveals exactly which model excels at what you actually do. More importantly, it exposes the overlooked feature (Projects) that can seriously multiply your AI productivity - and shows you how to use Deep Research to make better decisions faster.
AI Model Comparison: GPT-5, Claude & Gemini Showdown | Peter Yang
🎥 Watch the full episode here:
📆 Published: 25th October 2025
🕒 Estimated Reading Time: 3 mins. Time saved: 25+ mins! 🔥
Key insights from the full article:
🏆 The model matchup — GPT-5 for speed, Claude for writing/research, Gemini for images (with specific winners for coding, web search, video generation)
🧠 Projects: The secret weapon — attach Google Docs, templates, and context once, then iterate without repeating yourself each conversation
📋 Strategy documents in days, not weeks — upload templates + research + rough drafts, then use AI as your thought partner to iterate before sharing with stakeholders
🎬 Nine projects that save 3–10 hours weekly — show notes generator, family trip planner, product strategy hub, and more real-world examples
🔮 Deep Research mastery — five steps to get concise, personalised reports instead of 30-page data dumps (hint: context is the biggest lever)
🎯 The three-step framework — match models to tasks, build one project for your workflow, use Deep Research for decisions
💼 The compounding effect — each project gets smarter as you add context; within weeks, you've built a personalised AI system that understands your business
🗓️ Real examples that work — from finding the right piano app to planning multi-generational family trips with personalised recommendations
👉 Read the full breakdown — sent separately, check your inbox or click below
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 🍽️.