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? 🥘

OpenAI declares "code red" whilst users defect to Gemini (being first doesn't guarantee staying first), AI shopping jumps 805% on Black Friday (conversational commerce works it seems), and HSBC projects OpenAI needs another $207B before profitability (inconvenient maths). Meanwhile, Warner Music partners with the AI platform they were suing six months ago (fastest pivot in music history).

On the menu:

Oh hello, didn't we say don't discount Google ??? OpenAI's "code red" whilst users defect to Gemini proves being first doesn't guarantee staying first. Chinese models matching GPT-5 at a fraction of Western costs forces an uncomfortable question: what are we paying for? Runway beating Google and OpenAI at video proves foundation model advantages don't automatically transfer to specialised domains. Anthropic IPO prep, Apple's Gemini talent hire - the model wars are now about distribution, cost, and infrastructure control.

805% jump in AI shopping traffic is mainstream consumer behaviour. AI shoppers convert 16% better - conversational commerce just won the argument. OpenAI testing ChatGPT ads shows the real money's in commerce, not subscriptions. Amazon Rufus outperforming by 5x suggests they've cracked conversational interfaces that don't annoy people.

Amazon's $50B government deal whilst claiming 50% cost savings vs Nvidia is textbook platform strategy. HSBC's brutal $207B projection exposes the gap between valuations and economics. Warner partnering with Suno after suing them is the fastest "threat to partner" pivot in music history. SHIRLEY? Google possibly selling TPUs to Meta shows custom silicon's too expensive alone. Infrastructure consolidates whilst startups build on quicksand.

33% of high-performers scaling agents whilst others debate "will AI take jobs?" shows the gap continues to widen. Anthropic's 80% time savings measures task completion, not quality - convenient omission. Mistral's device-first bet is the contrarian play nobody else has courage for. Booking.com's guest messaging work is the boring enterprise case that'll scale faster than flashy demos.

Apple's crease-free foldable at $2,400 whilst Alibaba undercuts Meta by $262 splits hardware into premium innovation vs mass-market disruption. Samsung's trifold launching US early 2026 shows foldable form factors multiplying. Figure's "skull-fracturing" lawsuit is the robotics reality check the hype needed.

The year-end recap wave proves every platform's copying the same playbook - weaponised nostalgia drives engagement. Real innovation's happening at the edges: Discord monetising through embedded commerce, Netflix pivoting to mobile gaming, Healthify upgrading to real-time voice AI. MKBHD's Panels shutdown shows even 19 million YouTube subscribers doesn't guarantee app success.

Now through the end of the year is prime streaming time on Roku, with viewers spending 3.5 hours each day streaming content and shopping online. Roku Ads Manager simplifies campaign setup, lets you segment audiences, and provides real-time reporting. And, you can test creative variants and run shoppable ads to drive purchases directly on-screen.

Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

Use code GET5K now

Check the link here to access.

Tom takes us into the heart of Facebook's initial engineering culture, highlighting the groundbreaking approach of "Move fast and break things." He underscores the fine balance between speed and accountability, stressing that imperfection is an expected outcome of rapid innovation. Read the full article here.

💡 "If you never break anything, you’re probably not moving fast enough."

Craig Unsworth explores the rising significance of 'evals' in AI, structured assessments that critically evaluate model performance. As AI systems evolve rapidly, traditional assessments fall short; evals provide the structured measurement needed to ensure reliability and safety. Read the full article here.

💡 "Evals are not a nice-to-have. They are a continuous discipline." This quote encapsulates the critical nature of evals in ensuring AI systems function as intended and meet expected standards.

John Cutler explores the inherent differences in thinking styles—concrete versus abstract—and how they impact collaboration. He argues that these differences should not be seen as misalignment but as unique perspectives that can enrich teamwork if understood properly. Read the full article here.

💡 "We weren’t misaligned in intent. We just couldn’t hear each other through the noise of style." Understanding this quote reminds us that collaboration requires awareness and acceptance of differing cognitive styles for effective teamwork.

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.

🤖 📝 How To 10X Your Notes: Building Agentic Workflows with Claude Code and Obsidian

This started it all for me. McKay Wrigley's tutorial was the unlock—showing me how Obsidian + Claude Code creates an AI system that works with you, not just for you. Ten workflows that transform simple markdown files into an autonomous research assistant. This is where my 200-hour productivity journey began.

💡 Top tip - read the TL;DR below then shoot over to the full guide to deploy; if you really want to dig deep go straight to my my Claude Code series part 2 for even more

Key insights from the full article:

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 🍽️.