Thinking about hiring globally? Start with an EOR.
The best person for your next role might not live near your office—or even in the same country.
More companies are realizing they don't need to open entities everywhere just to access global talent. Instead, they're using EOR to hire internationally faster, stay compliant, and avoid building local infrastructure before they're ready.
Oyster's EOR helps companies hire, pay, and support employees in 180+ countries while Oyster handles payroll, compliance, taxes, and local employment requirements.

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? 🥘
Stripe spent the week trying to buy both the AI plumbing (OpenRouter, a cool $7B) and your actual wallet (PayPal talks are "heating up", apparently) - regulators presumably watching through their fingers. Anthropic casually mentioned a $65bn run rate while both frontier labs quietly paused their scariest training runs, which is a fun combination. Apple graciously opened the EU's walled garden and then stationed a man at the gate charging 15% on the way out. YouTube decided a "view" now starts the instant a video plays (every creator rate card just got a stealth raise), and a security researcher sketched out how a prompt injection worm would spread through all those shiny new agent-to-agent channels we shipped this month. Sleep well!
📰 Not Boring → Stripe's double acquisition play, the AI money league table, and platforms rewriting their own rules
⌚️ Productivity Tapas → Plain-English workflow automation, live session context for support agents, and more
🍔 Blog Bites → The $1.5M LinkedIn system, inside Canva's agentic era, and why a plan is just a thought
🎙️ Pod Shots → Thirteen conversations on the uneven 10x - who's actually getting faster with AI, and who's pretending
Let's go 🚀
📰 Not Boring
💰 Stripe Goes Shopping
Stripe reportedly agrees to acquire OpenRouter for more than $7B and the deal now looks clinched - a payments company buying the routing layer that sits between developers and every AI model
Stripe says "the singularity" has begun - Patrick Collison's framing for why a payments company needs an AI routing layer, in his own words
Stripe and Advent's talks to acquire PayPal are heating up - the same week as the OpenRouter deal, which would make Stripe both the AI toll booth and the consumer wallet
Workday shares surged on reported Silver Lake takeover talks - a 25% jump before trading was halted, as private equity circles enterprise SaaS
Kalshi is in talks to raise $750M at a $40B valuation while cracking down on insider trading - prediction markets maturing into regulated exchanges at venture speed
Jeff Dean's Disco Loop is in talks to raise $1B at $10B - the ex-Google chief scientist's startup covered here last week is already commanding a ten-figure round pre-product
Yuno raised a $45M Series B to scale its payments orchestration OS - orchestration keeps attracting money precisely because merchants refuse single-provider lock-in
Western Union got New York approval for its Intermax acquisition, with California hurdles remaining - state-by-state approval remains the tax on US payments M&A
Stripe buying OpenRouter and possibly PayPal in the same fortnight would put one private company across AI model routing, merchant checkout and consumer wallets. Regulators approved far smaller combinations slowly, so the sequencing of these two deals will matter as much as the price tags.
🤖 The AI Money League Table
Anthropic told investors its annualised revenue run rate hit $65 billion in July - up from roughly $9bn a year ago
OpenAI's run rate topped $40 billion ahead of its IPO but its Q2 sales growth looked tepid next to Anthropic's - the enterprise coding market is deciding the revenue race
OpenAI "will be a public company in 2027" or sooner, CFO Sarah Friar told employees - the IPO clock is now official and internal
SpaceXAI debuted Grok 4.6, overtaking Kimi K3 and matching GPT-5.6 Sol for third place on Artificial Analysis - benchmark third place is now a marketing event
Z.ai delayed its GLM-5.3 weights by two weeks after its cyber score beat Mythos 5 - an open-weights lab self-pausing on safety grounds is a first
DeepSeek publicised its push to challenge Claude Code and has passed Google to become the second-largest AI provider by token volume - volume leadership without revenue leadership is the Chinese playbook working
OpenAI previewed an Ultrafast API tier for GPT-5.6 Sol - latency is becoming a paid tier, not a given
OpenAI rewrote its safety framework while its largest training run stays paused and Anthropic says AI risks are rising with no plan to release a stronger "Model 2" - both frontier labs are now publicly throttling their own roadmaps
Two labs printing $105bn of combined run rate are simultaneously pausing training runs and shelving stronger models on risk grounds. Capability restraint and revenue growth are being managed as separate products, and for now the market is rewarding both.
🛠️ Agents Get an Org Chart
Separate Claude Code sessions can now message each other - coding agents are getting the coordination primitives of a team, not a tool
Linear's CPO says MCP usage is "through the roof" and Linear published data on AI usage patterns in software teams - agents reading the backlog directly is quietly replacing the status meeting
Confluence agents are now callable via @mention - Atlassian putting agents on the same footing as colleagues in a doc
How Ramp built third party product integrations that write themselves - integrations as an agent output rather than a roadmap line
One enterprise built an internal agent platform called Flux - the build-your-own-agent-platform pattern is spreading beyond big tech
How to manage AI agents effectively and an AI governance framework - management guidance for a workforce that does not attend the retro
I'm Worried About a Prompt Injection Worm - once agents message each other, a poisoned prompt can propagate like a virus
Introducing the Half-Day: 0-Day in the Age of AI - vulnerability disclosure windows collapsing from months to hours when agents do the exploiting
The same week agents gained inter-session messaging and @mention access to company docs, a security researcher mapped how a prompt injection worm would spread through exactly those channels. Agent connectivity is shipping faster than agent containment.
👩💻 Dev Platforms Reshuffle
Cursor launched Origin, its own code hosting platform and published Git at any scale - a direct run at GitHub while GitHub's reliability wobbles
Mojo is now open source - Modular following the 1.0 release covered last week by giving the language away to win the ecosystem
Foreman, Vercel's software factory template - packaged multi-agent software production you can clone off the shelf
GitHub vs Vercel vs Replit: what dev platforms do when AI code is cheap - when generation is free, platforms compete on hosting, review and deployment instead
AI is removing the middle class of software engineering and LinkedIn's internal AI code review system now handles 79,000 reviews a week - review capacity, not writing capacity, is the new bottleneck being automated
You can just choose how many bugs you want now - quality as a dial you set with review effort, not an accident of talent
AI automation startup Relay shut down, with staff joining Google's Chrome team - browser-level automation talent is consolidating into the platforms themselves
Cursor hosting code, Vercel templating factories and LinkedIn automating 79,000 reviews a week all point the same way: the value in developer tooling is moving from writing code to owning the pipeline around it.
📱 Platform Rules Rewritten
Apple overhauled its EU App Store fees and loosened rules for alternative app stores while proposing a 15% cut of purchases made outside the App Store - conceding the walls while re-pricing the exits
YouTube will now count a view as soon as a video starts playing - every creator dashboard and sponsorship rate card just inflated overnight
YouTube is offering creators millions not to work with Netflix - exclusivity payments arriving the same week the view metric got redefined
Reddit began testing a TikTok-style audio and video experience - the last big text platform hedging into short video
Spotify's new Playlist Notes let users and editors explain their song picks and it launched an "AI persona" badge identifying artists created or represented by AI - human curation and human provenance becoming labelled features
Anthropic shipped an invisible watermark marking content generated by Claude - AI provenance moving from voluntary badge to embedded default
Facebook rolled out its stand-alone Creator Studio app with AI tools for creators - unbundling the creator toolkit to compete for supply
X open sourced its ranking algorithm, letting users see if they've been 'shadowbanned' - algorithmic transparency as a competitive feature rather than a regulatory demand
Bluesky introduced Protocol Services - monetising the protocol layer instead of the feed
Shopify pushed back on the narrative that AI is cannibalising organic search - its merchants are seeing AI referral traffic convert, not evaporate
Apple loosening EU store rules while claiming 15% of external purchases shows how platform concessions actually work: the rule changes, the take rate survives. YouTube redefining a view days later is the same move in metrics form - platforms rewrite the units everyone else's business is measured in.
💳 Payments Plumbing Gets Crowded
Klarna went live with JP Morgan Payments for US merchants - BNPL distributed through the biggest acquirer in the US
Klarna's Q2 showed growth turning into profitability but its shares dipped on revised guidance - profitable and punished in the same earnings cycle
Visa and Mastercard are backing a new Agentic Payments Alliance - the schemes moving to own the standards before agent checkout ossifies without them
Adyen expanded its Toast partnership into the US - vertical software platforms keep picking one global processor rather than building their own
Airwallex partnered with Affirm for a US BNPL roll out - credit at checkout is now table stakes for cross-border players too
Monzo topped one million business accounts - the UK SME banking market is being won on onboarding speed, though chair Gary Hoffman is stepping down after a shareholder revolt
TikTok is exploring peer-to-peer payments via DMs and Venmo now takes college tuition payments - social platforms and P2P apps both stretching into money movement's edges
Robinhood brought zero-fee digital asset trading to the UK and its second private markets fund started trading - loss-leader trading funding a push into previously gated asset classes
Nvidia is the world's largest fintech company and the great unbundling of consumer fintech - two essays arguing the fintech category boundary has stopped meaning anything
Klarna inside JP Morgan, Adyen inside Toast, Affirm inside Airwallex: the pattern this week is distribution deals, not product launches. Payments capabilities are commoditising fast enough that owning the merchant or user relationship beats owning the rails.
📐 Pricing, Proxies & Benchmarks
"Your benchmarks don't apply to us" and The benchmarkpocalypse - two takedowns of comparing your team to industry numbers that were gamed before you read them
TBM 437: Tokens, Hours, Points, and Other Curious Proxies - John Cutler on token spend joining story points in the proxy-metric hall of shame
Input-based pricing vs output-based pricing and Peacock raising prices across all streaming plans - pricing theory and pricing practice in the same week
Cracks in the AI Thesis - Ramp's August AI index - corporate card data showing AI spend growth flattening for the first time
Rethinking the Data Moat and Why you can't copy Palantir - both land on the same conclusion, that moats live in workflow entanglement rather than data volume
9 theses on AI - a compact set of positions worth stealing for your next strategy offsite
Evaluating AI Agents as Products - a long, practical framework for judging agent quality beyond demo theatre
Ramp's card data showing AI spend flattening landed the same week as two essays dismantling benchmark culture. Teams that justified AI budgets with borrowed benchmarks now need their own product-level evidence, and the eval frameworks to produce it are finally being written.
🌍 Distribution Land-Grabs
Perplexity's free AI offer left it with millions more users in India - buying market share in the largest open AI market before defaults harden
Google Pay introduced a Gemini-enabled AI tool for users in India - the same battleground, fought through payments rather than search
Apple Pay is advancing launch plans for India and Iraq - a third front in the same land-grab, this time at the wallet layer
OpenAI launched a safer ChatGPT for teens while US states seek $200 billion from Meta over child social media addiction claims - youth safety products shipping in the shadow of youth safety litigation, a week after Meta's $942m ruling
Calendly threw its hat into the meeting note-taker circus - scheduling tools chasing the meeting-intelligence wedge that Granola and Zoom already occupy
Google bought a load of Spirit Airlines data for AI training - distressed companies discovering their operational data is an asset class
OpenAI instituted new safeguards after the Hugging Face breach - the formal policy response to the incident covered over the last two issues
Sundar Pichai directly urged key AI staff to prioritise Gemini model development - post-Jeff Dean Google concentrating its bets rather than spreading them
India is this week's clearest prize: Perplexity is giving its product away there while Google routes Gemini through the payments app Indians already use daily. Free access wins the install, but the payments integration wins the habit.
⌚️ Productivity Tapas: Time-Saving Tools & Workflow Automation
Ballet: Describe a workflow in plain English and it writes deterministic automation code with audit logs and rollback - built for RevOps, support ops and PLG teams who've outgrown Zapier-style brittleness.
Fullstory for Customer Agents: Streams real-time session context (what the user actually did) into your support platform so AI and human agents stop asking customers to describe their screen. GA now.
Framer Agents: Framer's canvas agents just got a reasoning picker - light reasoning for quick edits, deep for planning whole builds. Worth a look if marketing site iteration is stuck in your backlog.
Sidekick: An agentic command layer for macOS that carries out tasks across your apps and files from natural language. Free beta with $0.50 of credits, $5 top-ups.
Deltix: Describe a test in plain English and an AI agent runs it against your iOS app on a simulator - user-workflow validation without writing a line of test code. Free in open beta.
Remember. Product Tapas subscribers get our complete toolkit - 580+ 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
Growth: The Five-Pillar LinkedIn System Behind a $1.5M Business
Will from The Growth Playbook breaks down how Exit Five, a media company for B2B marketers, built a business reported at north of $1.5M a year with LinkedIn as the core engine - and why its audience is smaller than you'd guess. Read the full article here.
💡 "A tight audience of buyers beats a huge audience of strangers every time."
Key Takeaways
Pick a slot in your buyer's head: Exit Five's positioning is aggressively narrow - the go-to resource for B2B marketers. You don't need a massive following: 5,000 to 7,000 highly relevant, engaged followers can support a real business, as long as they're the right people.
Run a funnel on your feed: Mix three levels deliberately - broad relatable content for reach, tactical B2B playbooks as the bulk of output, and direct promotion where attention turns into revenue. The mistake is living at one level; the middle is where compounding trust lives.
Build a content ecosystem, not a personal brand: Exit Five runs a cast - founder, COO, head of content, marketing manager, ops, plus the company page. More reach, omnipresence in the feed, better ICP coverage, and a recruiting magnet all follow.
One long-form asset becomes many posts: The podcast fuels the whole feed - each episode yields five to ten clips distributed across team accounts, the company page, and the guest, who gets clips and suggested copy so sharing is effortless.
The real moat is time: The results come from years of near-daily consistency, not a 30-day sprint. Commit to 12 to 24 months and measure leading indicators first - posting volume, profile views, DMs, saves, follower quality. Revenue lags, then compounds.
Will, The Growth Playbook
Why I picked this: A rare LinkedIn playbook that's an actual system rather than hustle-bro tips - the ecosystem pillar (a cast of posters, not one founder account) is the piece most product-led companies could steal tomorrow.
Data & AI: Inside Canva's Agentic Era: One AI Product for 265 Million Users
Bill Kerr interviews Danny Wu, Head of AI Products at Canva - the $42 billion design platform with 265 million monthly active users - on turning a pile of separate AI features into one conversational, agentic product. Read the full article here.
💡 "What's really exciting is that as visual mediums become more accessible, the average person is able to show what they mean rather than trying to describe it. Visuals will become the new shorthand."
Key Takeaways
The glue role: Wu isn't part of any single AI team - he sits across Canva's supergroups as connective tissue, stitching dozens of teams with shared dependencies into one coherent product. His warning: with AI making prototyping cheap, it's easier than ever to "ship your org structure".
From tool to agent: Canva AI pulls context from where work already happens (Slack, email), then starts the design for you and with you. A core design constraint: AI outputs are real, hand-editable designs, not images you must re-prompt to change.
Six-week sprints, low bar for ideas: Most AI teams run six-week sprints - not to limit ambition but to coordinate shared ownership. Ideas start as quick wireframes circulated for traction; much of the roadmap is bottom-up.
Small models, specialised training: Design-model work starts with tiny toy-model experiments to find what moves benchmarks per training step. The philosophy: rely on general-purpose LLMs for general tasks, and build a model "really, really good at the specialized thing".
AI-free days: Wu's personal curveball - regular days using no AI at all, to keep the cognitive muscles working that easy AI assistance quietly atrophies.
Bill Kerr, Open Source CEO
Why I picked this: A genuinely detailed look at how a large product org operationalises AI - org design, cadence, evals and training included. The "ship your org structure" warning and the AI-free days habit are both keepers.
Strategy: A Plan Is Just a Thought: Planning When Reality Pushes Back
Tom and Corissa at Crown & Reach share a batch of hand-picked quotes on planning in complexity - trying to control the uncontrollable - alongside five recent episodes of their Tentacles podcast on strategy and knowledge creation. Read the full article here.
💡 "We treat our plans as though they are a lasso, thrown from the present around the future, in order to bring it under our command."
Key Takeaways
Change pushes back: Adrian Tchaikovsky's Cage of Souls captures it: "when you set out to change things, the things you want to change push back." Worth pinning above any transformation roadmap.
The vision myth: Richard Rumelt (Good Strategy Bad Strategy) argues that ascribing Ford and Apple's success to a shared vision, rather than pockets of outstanding competence mixed with luck, is "a radical distortion of history".
Plans are statements of intent: Oliver Burkeman's Four Thousand Weeks reminds us "a plan is just a thought" - a present-moment expression of how you'd like to deploy your modest influence. The future is under no obligation to comply.
Beefs with backcasting: Tentacles episode 148 gives away a complete two-day strategy workshop designed to feel like backcasting while breaking free of the problems backcasting induces; episode 144 asks whether strategy, as everyone sells it, even exists.
Tom & Corissa, Crown & Reach
Why I picked this: Three short quotes that puncture more strategy theatre than most full-length books. If your quarterly plan keeps colliding with reality, this reframe - plans as intent, not lassos - is a healthier way to hold it.
🎙️ 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
Pod Shots #149 - The Uneven 10x
Ian Silber, head of design at OpenAI, has cross-workforce survey data showing designers and user researchers are the most overwhelmed, most anxious, least optimistic and most tired cohort in tech, and he names the cause himself: engineering got 100x faster while the design process still takes the time it takes. Klaviyo's CEO, Travis Kalanick, Susan Kare and a growth lead who scaled Wispr Flow converge on the consequence - the constraint has moved from hands to attention - and this brief works out what that means for staffing, spend and sequencing.
Featured in this round-up:
🎙️ a16z Podcast - "Ben Horowitz and Travis Kalanick on Building Again" and Founders Podcast - "Excellence Is the Capacity To Take Pain" with Travis Kalanick - merged as one perspective
🎙️ 20VC - "How to Build a $100M Growth Engine" with Matt Swulinski
🎙️ SaaStr - "Agents Are Your New Power Users" with Andrew Bialecki
🎙️ Y Combinator Lightcone - "Designing Icons & Graphics For the Original Mac" with Susan Kare
🎙️ Latent Space - "The BioAI Phase Shift" with Matthew McPartlon and Neil Patil
🎙️ The Pragmatic Engineer - "Stop being skeptical about AI for development" with Charity Majors
🎙️ Possible - "Making taxes fun with Pikachu and AI" with Cadi Zhang
🎙️ The Diary of a CEO - "A Kid In The 29th Century Could Have Made This World" with Prof Brian Greene
🎙️ Decoder - "Does Google even want to win in AI?"
🎙️ Big Technology - "How Ozempic Changes Our Bodies, Minds, and Economy" with Johann Hari
🕒 Estimated Reading Time: 10 mins. Time saved: 748 mins! 🔥
🚪 Not feeling this one? Five from the archive that might land better:
Pod Shots #127 - Public Speaking - how to speak clearly and with confidence, no agents required.
Pod Shots #123 - Netflix's Culture - Elizabeth Stone on talent density, candour and what it costs to sustain both.
Pod Shots #117 - Pricing - the 4x growth lever most companies leave sitting on the table.
Pod Shots #131 - Global Supply Chains and AI - the 21km gap that could crash the global economy.
Pod Shots #119 - High Trust Teams - difficult conversations and how the best teams design for them.
Key insights from the full round-up:
📉 The unhappiest people in tech are designers - bottom on every dimension of Silber's survey; the mechanism is an unevenly distributed productivity gain.
🧭 Management capacity is the new ceiling - four shows land on it independently. You can hire out of a capacity problem, not an attention problem.
🎨 Taste compounds when output is free - Kare and Zhang, forty years apart: machines get you to approximate quickly and stall at exact.
🏆 The rare talent premium is widening fast - marketing acquisition gets its ML-researcher pay moment within a year.
💰 The e-com playbook is the SaaS playbook - switch paid ads on in week one as your fastest test instrument, not your growth channel.
🤖 Agents are users, not tools - Klaviyo asks its agents what is blocking them, and reopened a feature humans couldn't use for a decade.
🔬 Frontier models read about science rather than do it - the observations that matter sit in raw data nobody wrote up.
📈 Value is migrating faster than attention - Airtable $11.7bn to $1.29bn; GLP-1 revenue outran every AI lab combined.
🌊 Both AI camps are right at the same time - Majors lands on the thing nobody budgets for: worse before better.
🔧 The capacity to take pain, revised - Kalanick concedes he ran a $70bn company like a founder who might not make rent.
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 🍽️.

