How I AI – "How Coinbase scaled AI to 1,000+ engineers"
Host: Claire Vo | Guest: Chintan Turakhia (Senior Director of Engineering @ Coinbase)
Date: March 2, 2026

Episode Overview

In this episode, Claire Vo sits down with Chintan Turakhia to explore how Coinbase successfully drove AI adoption across a massive engineering organization. Chintan offers a practical, tactical, and deeply candid view of leading AI transformation for 1,000+ engineers, unpacking specific workflows, real-world wins, common mistakes, and the new cultural expectations AI brings to engineering management and teams. In true How I AI fashion, the episode is full of actionable tips, technical walk-throughs, and even a personal “life hack” in the final stretch.

Key Discussion Points & Insights

1. The Reality of Scaling AI at Enterprise-Level Engineering

Timestamps: [00:00]–[04:00]

• Skepticism persists on whether large, capable engineering orgs can really adopt AI at scale and win tangible efficiency gains.
• Chintan’s assertion: “It's not only possible, it's adapt or die. It's just been such a huge superpower for the team.” [00:11, 03:10]
• Adoption numbers: Coinbase embedded AI into the workflows of 1,000+ engineers. [00:17, 03:56]

2. The Path from Failed Pilots to Real Transformation

Timestamps: [04:00]–[08:00]

• Early AI tool efforts (Cursor, Copilot) met with mixed results; initial trials left many engineers cold, creating organizational skepticism.
• Leadership lesson: Organizational transformation requires “a single person with incredible conviction at the leadership level who is also hands on the metal.” (Claire Vo, [04:34], also restated at [08:24])
• “The worst thing any eng leader could do is just be like, I decree you must use AI. Come on. No one's going to listen to you.” (Chintan, [04:45])

3. Making AI Adoption Stick: Show, Don’t Tell

Timestamps: [08:00]–[11:50]

• Chintan personally dogfooded AI tools “every single day, every single hour,” using Cursor for code, paperwork, and recruitment.
• Importance of demonstrating small, tangible wins: “The best way to get to an engineer is just give them the tools so they stop doing the shit work and so they can build the stuff they love.” [09:57]
• Focused on using AI for toil: unit tests, linting, and basic bug fixes—targeting “paper cuts” that slow engineers down.
• Aha moment: Automating PRs and “never need to remember git status, git rebase, not like, why is anyone doing this anymore?” [10:30]

4. Building Momentum: Social Proof and Speed Runs

Timestamps: [11:50]–[15:30]

• Created a “cursor wins” Slack channel: engineers posted time savings and AI achievements, driving internal visibility and peer conviction. [11:33]
• Introduced the team-wide PR Speedrun: with 100 engineers submitting 70 PRs in 15 minutes using AI tools, “We broke GitHub too, which was cool... our infrastructure needed improvement.” [12:53]
• Cultural shift: moving from rigid approval chains to radical ownership and increased shipping velocity.
• “What I really like... This is a moment where we should be breaking the rules, because AI is breaking the rules for us.” [15:32]

5. Metrics and Measuring Impact

Timestamps: [15:30]–[19:30]

• “How do you measure any of this, like, in terms of output?” Not as a replacement for people, but as an accelerant.
• The key metric: Time from ticket → user impact.
  • Ticket creation → PR → code review → merge → release.
  • “We reduced PR review cycle time by 10x, down to ~15 hours from 150.” [17:21]
• AI automation (“agent picks it up”) compresses feedback loops dramatically; real-world example: shipping a requested change live to a user during a 30-min call. [18:56]

6. Quantitative AI Usage Analysis and Cohort Identification

Timestamps: [19:30]–[34:00]

• Using Cursor admin analytics CSVs, Chintan demoed generating enriched analyses to:
  • Identify usage clusters: agent-heavy, tab-heavy, balanced, and non-users (“cursor-pilled/lite”).
  • Share actionable guidance and cohort-based playbooks (with motivational Slack snippets).
• “Stop typing. Start shipping.” (LLM-generated slogan, [32:14])
• Fun “gamified” playbook outputs: e.g., “tab harder” (potential internal merch!). [32:39, 33:08]
• Visualizations and personalized feedback fostered virality and engagement among engineers.

7. Supercharging Feedback Loops with Custom Bots

Timestamps: [36:28]–[47:09]

• Chintan built a live audio/video feedback tool integrating LLMs:
  • Captured unstructured feedback, summarized bug reports, generated Linear tickets, and triggered agents to author PRs—within minutes.
  • “Somebody manually summarizing... would have been a huge bottleneck in the past.”
• Details on Coinbase’s custom Slackbot “cloudbot,” plan mode agents, and strategies for integrating with Linear, Sentry, Datadog, and more, ensuring rich internal context for AI-powered code/gen.
• “The trick here. Why Slack? Because Slack is how things go viral within your company... if you have pulled out the magic into some separate tool that others can't see, it doesn't happen.” [49:37]

8. Building the "Super Builder" Role and Culture

Timestamps: [43:12]–[49:01]

• Coined and hired for the “Super Builder” role:
  • “The single most important job of a super builder is to create more super builders.”
• Advice from Claire: being a top AI-pilled engineer is massive career leverage: “You want to be like the top three most AI pilled people in your engineering organization... It is incredible benefit to your overall career.” [47:09]
• Reinforced: “Building your own agents is not impossible for organizations,” especially with security/compliance needs.

9. Personal AI "Life Hacks": Beyond Engineering

Timestamps: [50:14]–[55:24]

• Time savings: parse school emails by snapping a photo and letting ChatGPT generate calendar invites.
• Personal Champagne Sommelier:
  • Chintan reverse-engineered his wine preferences from tasting notes, then used those to get recommendations from restaurant wine lists by snapping photos and prompting: “What would I like from this list? What are good values?”
  • Technique mirrors earlier guests (Whoop, photo styles) but first wine-specific use on the show.

10. Leadership Reflections & Lightning Round

Timestamps: [55:24]–[57:27]

• AI impact on work style:
  • “My calendar's empty—almost empty. The coordination overhead of, like, hey, let's prioritize this... No, you just do things... I'm writing way more code.” [55:42]
• On dealing with LLMs that fail to follow instructions:
  • “Okay, one, you're clearly not listening to me. This is what I said... I threaten it and I say... I'm going to switch to Gemini and then it gets a chip together.” [56:43]

Notable Quotes & Memorable Moments

• Chintan Turakhia:

  • “It's not only possible, it's adapt or die.” [00:11, 03:10]
  • “Show the engineers, not just tell. The worst thing any eng leader could do is just be like, I decree you must use AI.” [04:45, 09:57]
  • “Stop typing. Start shipping.” [32:14, LLM-generated]
  • On PR Speedruns: “In 15 minutes we ended up putting up like 70 PRs. And we broke GitHub too, which was cool.” [12:53]
  • “AI is an accelerant because there will always be more work to do.” [16:09]
  • “My calendar's empty—almost empty. The coordination overhead... no, you just do things. I'm writing way more code.” [55:42]
  • “If Cursor is listening, I think this is going to be your new merch line: ‘tab harder.’” [33:08]
  • On prompting stubborn LLMs: “I threaten it and I say... I'm going to switch to Gemini and then it gets a chip together.” [56:43]

• Claire Vo:

  • “Organizational transformation requires a single person with incredible conviction at the leadership level who is also hands on the metal.” [04:34; echoed at 08:24]
  • “No one's getting bonus points for memorizing git commands.” [11:30]
  • On culture: “Breaking the rules is so powerful for velocity and for... radical ownership of things.” [14:41]
  • “You want to be like the top three most AI pilled people in your engineering organization.” [47:09]
  • “If you are thinking about driving AI adoption in your org, figure out how to get the right platforms in place that can unlock access to agents... If you ask somebody to open or learn a new tool, it's just going to create too much friction.” [48:47—paraphrased]

Important Segments & Timestamps

• [03:56] – Chintan shares the context & scale of the AI rollout (1,000+ engineers)
• [10:30] – Automating PR creation and the “aha” moment
• [12:53] – Team PR Speedrun in 15 minutes: shipping at scale, “broke GitHub”
• [17:21] – 10x reduction in PR cycle time using AI agents
• [21:59] – Using Cursor for analytics and cohort analysis of adoption
• [32:39] – Playbook gamification: “tab harder”, motivational prompts for user cohorting
• [36:28] – Demo: feedback pipeline from audio to PR via AI and in-house bots
• [47:09] – Role of “super builder” and cultural drivers
• [55:42] – How AI changed Chintan’s leadership style and time use

Actionable Takeaways

• Hands-on, visible leadership is essential for AI transformation—lead by example.
• Start AI adoption with “toil” (testing, linting) and tangible wins to build internal momentum.
• Gamified, transparent Slack channels (sharing wins/losses) create social proof and peer learning.
• Time-bound “speedruns” and all-hands exercises surface proof points and break rules that block progress.
• Leverage LLMs for analytics, playbooks, and strategy—automating what used to be time-consuming management busywork.
• Custom bots and agents, especially integrated into Slack and Linear, eliminate friction and supercharge feedback-to-feature cycles.
• Building in-house agents is feasible and advisable for organizations with high compliance/security needs.
• Embrace a culture of creation: collapse meeting time, maximize direct code contribution, and empower teams to self-serve.
• “Super builder” roles—engineers dedicated to disseminating AI skills—can multiply org impact and are a top career opportunity.

For Listeners

• Connect with Chintan at @ChintanTarakhia (Twitter/X)
• Check out the new Coinbase “base app”—launch imminent; feedback and bug reports welcome!
• Interested in “super builder” roles or helping scale AI in engineering? Coinbase is hiring.

Summary prepared for those seeking pragmatic insights into organizational AI adoption, internal tool-building, and the culture shift required to ship faster and smarter with AI. Highly recommended for engineering managers, product leaders, and technical ICs at scale.