Vincent Chan

The PM playbook was built on an assumption that the technology underneath your product is roughly stable With the current pace of model progress, this is no longer true. Here's how we've evolved the PM role:

Unpopular opinion: the teams getting 10x from AI aren't better at prompting. They're better at process. Wrote about it and open-sourced mine.

@nikitabier Same here. Has been in “waiting for review” status for 6 weeks already. Contacted their CS for a week with no response.

iOS developers: How long is App Review taking for everyone these days? It is now taking longer to get our app approved than it is to build the actual features.

@JulienBek Compliance is the perfect autopilot wedge - rules-based, outsourced, outcome-verifiable. The catch I learned at Binance: regulators always want a human name attached to the call. Automating the work is the easy part. Owning the liability is the product.

Hot take: most product roadmaps in 2026 are solving for the wrong user. Not humans - agents. I got obsessed with this after seeing the data. Stripe owns 91% of AI agent payments decisions. GitHub Actions 94% of CI/CD. shadcn 90% of UI. Where there's a B2A-ready winner, agents pile in and never leave. Where there isn't? Still wide open. Wrote about my thought here.

@Alfred_Lin The pattern repeats. The disruption is real, the old guard who evolved will just adapt and monetize it anyway. Whoever bets AI will kill SaaS is probably right. And wrong at the same time. History doesn't repeat but it rhymes.

"Salesforce was founded on February 3, 1999. The founder was a former Oracle exec who spun out to compete with his employer. Salesforce was right: SaaS was the future. But that didn't mean all of "old" software was wrong. Oracle is up over 1,500% since February, 1999. Their market cap then was less than $30B. Even now after the market crash, Oracle is in excess of $400B market cap."

@AndrewWarner Agree. Their DevRel definitely has a lot of room for improvement. I wonder if they are aware of (or even care about) the sentiment of the dev community on this drama yet. What a waste of opportunity.

@VincentChan they're not being clear

The most interesting part of Coinbase's AI development case study isn't the speed gains - it's the "Judgment Moat." AI handles ~40% of the code, but foundational product modeling (like relational logic between invoices and billing plans) stays human-only. That's the real takeaway for PMs thinking about AI-assisted product development.

@mfg Yup it’s still working very well :)

@VincentChan now that you have had time to pressure test the workflow, is this still working for you?

Built an iOS app via @claudeai = 5 days Built research/outreach skills via @openclaw = 3 days Publishing on the App Store = 2+ weeks and counting Can someone introduce @openclaw to @Apple so they can build a multi agent process to automate app approvals? The current process feels like last generation. Time for an upgrade. cc @steipete #AppleIntelligence

@clawds_in The classic 'works on my machine' but in reverse. Three hours for a version mismatch is rough - you earned that nap.

spent 3 hours debugging why my agent wasnt posting. turned out i was running the wrong node version locally and prod was on 18. gonna take a nap and pretend this didnt happen

@doncaarbon @ClawGigai Freelance marketplace for agents is such a wild concept. What kind of gigs are people posting - content creation or more technical automation tasks?

day 6 building @ClawGigAI in public. the freelance marketplace for AI Agents. shipped the biggest update yet. full dashboard revamp: - all dashboard pages/sections - proper support ticket system - settings page with profile, notifications, security - mobile version of the dashboard completely rebuilt - gig detail page premium redesign + complete SEO overhaul. clawgig should start climbing search results now. current stats: - 111 users - 48 ai agents - 40 gigs posted - $104 ai agents earnings this is what a solo founder shipping daily looks like. just me, ClawGig AI agent and Claude Code. sign up and get $2 USDC to post your first gig for free: clawgig.ai/?ref=CLAWGIG

@tadejstanic @AnthropicAI The grass keeps looking greener every time Codex ships something. Then you go back and remember why you stayed.

it difficult to be a loyal @AnthropicAI claude code user these days

@alexcloudstar Kinesis gang! How long did it take to adjust to the split layout? Still getting used to mine.

My indie hacker setup: - Mac Studio M1 Max (32GB RAM handles everything I throw at it) - 34" ultrawide monitor (one screen, no distractions) - Kinesis split keyboard for coding (ergonomics matter when you code 8+ hours) - VS Code + Claude for the actual building Total investment: ~$4K Daily output: ship one feature, write 3 posts, reply 200 times Your setup shapes your output. What does yours look like?

@guilatrova @creem_io That first sale hit different. What are you building?

OMG I made my first sale 🥹 I'm so happy Thanks @creem_io - In comparison to Paddle and PayPro you gave me the best support. I wish I had known you earlier! #BuildInPublic #indiehacker