Skip to content
Aakash GuptaAakash Gupta

How to Build a Company OS in Claude Code

Jiaona Zhang (JZ) is the CPO at Laurel, a $100M AI timesheet platform. She has led product at Airbnb, Dropbox, Webflow, and WeWork. Today she runs a product team that ships front-end and back-end features end-to-end. In this episode, she screenshares Laurel's full Company OS live, walks through the agent pipeline, shows how non-technical team members ship to production using AI, and breaks down the 4 levels of AI maturity she uses to assess every candidate she interviews. Full Writeup: https://www.news.aakashg.com/p/how-to-build-an-ai-native-team Transcript: https://www.aakashg.com/how-to-build-ai-native-team/ Laurel: https://www.laurel.ai/ --- Timestamps: 0:00 - Intro 1:46 - Episode begins 2:04 - The Company OS: GitHub structure screenshare 5:40 - The 1% vs 99% problem 9:00 - 3 steps to build your own Company OS 10:05 - Ads 12:30 - Slack automation demo: feature request triage 14:31 - Playbook to agent pipeline 22:51 - Company culture and the companywide hackathon 29:02 - PMs shipping front-end and back-end features 29:44 - The captain model explained 30:34 - Ads 32:37 - Continuation to captain model 37:38 - Two-track product reviews 50:08 - The AI Ops team and the Sasha model 57:59 - The screen-share interview 59:01 - The 4 levels of AI maturity 1:06:08 - Outro --- 🏆 Thanks to our sponsors: 1. Ariso - Ship AI agents and features faster, with fewer regressions - https://ariso.ai/aakash 2. Bolt - Ship AI-powered products 10x faster - https://bolt.new/solutions/product-manager?utm_source=Promoted&utm_medium=email&utm_campaign=aakash-product-growth 3. Pendo - The #1 software experience management platform - http://www.pendo.io/aakash 4. Product Faculty - Get $550 off their #1 AI PM Certification with code AAKASH550C7 - https://maven.com/product-faculty/ai-product-management-certification?promoCode=AAKASH550C7 5. Customer.io - Send smarter messages using your product data - http://customer.io/productgrowth --- Key Takeaways: 1. Every company has a 1% who are AI-native and a 99% who do not know what to use when. The Company OS closes that gap by encoding the 1%'s workflows into skills that anyone can use when they open Claude. 2. Build the ontology before you build the OS. Map every team's work to categories and tasks first. Color-code what should get more human time vs what gets automated. The OS is built from that work map. 3. Even the friction of going to a different interface kills adoption. A separate agent tool in a new tab will not get used consistently. Deliver skills and automations inside Slack and email, where people already are. 4. When AI adoption is everyone's responsibility, it is no one's responsibility. Dedicate one person full-time to AI Operations. Start with one person who demonstrates value. Every other function will want their own version within months. 5. The Company OS turns a 50-page playbook into a set of agents. Write the playbook first. Then audit it. What requires a human? What can be automated? Build the skill files from what remains. 6. The captain model replaces the handoff chain. Every feature has one owner end-to-end. The captain is whoever has the most critical skill for that feature's hardest problem. 7. PMs at Laurel ship front-end and back-end features. Not just growth experiments or copy changes. Core product features deeply integrated with billing systems and time entry logic. One PM who identifies as a designer shipped one of these end-to-end last month. 8. JZ went from hundreds of reports to 5 PMs and 4 designers. They ship more than ever. Adding people adds coordination cost. In a world where one PM can take a feature from discovery to production in a day, large teams cancel out their own capacity gains. 9. The new PM interview is a screen share. JZ asks every candidate to show their actual screen. In 60 seconds she knows their level of AI skills. 10. The PM fundamentals never changed. Problem space first. Know why and for whom you are building before you build. The speed changed dramatically. What you are supposed to be doing at the heart of it did not. --- 👨‍💻 Where to find Jiaona Zhang: LinkedIn: https://www.linkedin.com/in/jiaona/ Reforge: https://www.reforge.com/profiles/jiaona-zhang Laurel: https://www.laurel.ai/ 👨‍💻 Where to find Aakash: X: https://x.com/aakashgupta LinkedIn: https://www.linkedin.com/in/aagupta/ Newsletter: https://www.news.aakashg.com #productmanagement #aipm #claude --- About Product Growth: The world's largest podcast focused solely on product + growth, with over 200K+ listeners. Subscribe and turn on notifications.

Jiaona ZhangguestAakash Guptahost
Jun 24, 20261h 7mWatch on YouTube ↗

CHAPTERS

  1. Why Laurel’s CPO interviews for “AI-pilled” product builders

    Aakash introduces Jiaona “JZ” Zhang and frames the episode around a provocative idea: the most effective AI-native PMs operate like end-to-end builders. JZ previews how fundamentals stay the same while tools and operating systems change how teams execute.

    • JZ’s background (Laurel CPO; ex-Linktree, Airbnb, Dropbox, Webflow, WeWork; Stanford instructor)
    • What’s changing: PM execution model vs. PM principles
    • Teaser: JZ’s interview tactic to verify real AI fluency
    • Preview of AI maturity levels and company operating systems
  2. Inside Laurel’s “Company OS” in GitHub: functions, folders, and skills

    JZ screen-shares Laurel’s GitHub repo that acts as a companywide operating system, organized by function and phases of work. The key concept is turning repeated work into reusable “skills” that are discoverable and standardized across the org.

    • GitHub as the source of truth: folders for CS, sales, marketing, design, engineering, finance, etc.
    • Each workflow area contains playbooks/skill docs (e.g., renewals, negotiation support, onboarding)
    • Skills are meant to be executed repeatedly, not just documented
    • The OS maps real work, not abstract org charts
  3. Making the OS real: Slack-driven daily briefings + Claude skills

    JZ shows how the OS gets embedded into daily work via Slack, including a daily briefing pattern. Skills are uploaded into AI tools (e.g., Claude) so people can call the right asset at the right moment instead of reinventing decks/emails.

    • Most work happens in Slack/email—delivery must meet people where they work
    • Daily briefings: calendar + meeting prep + handoff/session prep skills
    • Claude ‘skills’ as company context to reduce repeated content creation
    • Operational goal: consistent customer experience through shared motions
  4. The “1% vs 99%” AI adoption gap—and how a Company OS fixes it

    JZ explains the common adoption failure: a small group of power users builds great workflows, while most employees don’t know what tool/skill to use when. The OS converts individual experimentation into organization-wide leverage.

    • 1% of power users vs. the rest of the org lacking clarity
    • Standardizing ‘what to use when’ across roles and functions
    • Onboarding people into shared skills instead of ad hoc tips
    • Scaling best practices from top performers to everyone
  5. Step 1 of building your own OS: automate one painful workflow (feature request triage)

    JZ outlines a three-step path to build an operating system, starting with a single annoying workflow. She demos a Slack automation that structures and routes feature requests to reduce back-and-forth and improve triage speed.

    • Pick one repetitive task you’d love to never do manually again
    • Slack workflow collects required context (impact, frequency, recordings, details)
    • Auto-routing to the right owner + creating a trackable ticket
    • Define SLAs and reduce ambiguity in intake processes
  6. Step 2: from playbooks to agent pipelines (and why a ‘mega-agent’ matters)

    Next, JZ shows how large playbooks become usable by translating steps into agents and automations. The breakthrough is creating a single ‘mega-agent’ interface that routes requests to the right sub-agent so teams don’t need to remember dozens of tools.

    • Playbooks can be long (50+ pages) but must translate into action
    • Audit steps: human-required vs. automatable/productizable steps
    • Agent builders (Dust/Glean/Claude/OpenAI) can encode sub-workflows
    • Mega-agent pattern: one entry point that routes to specialized agents
    • Adoption hinges on minimizing interface switching; deliver in Slack/email
  7. Tooling strategy: Dust vs. Claude + avoiding automation overload

    JZ explains why Laurel initially used Dust for agent building and how that gap is shrinking as Claude improves. She also warns that scheduled automations can overwhelm teams, motivating consolidation into a single OS with curated, high-signal delivery.

    • Early tooling maturity favored dedicated agent builders; now Claude can cover more
    • Skills can be stored as files and invoked directly inside Claude
    • Scheduled tasks can become noise; information overload reduces adoption
    • Company-level curation creates consistency across functions
  8. Culture as the multiplier: hackathons, ‘everyone is a builder,’ and enabling non-technical shipping

    JZ argues the OS only works with the right culture—leadership must signal it’s cross-company, not just engineering. She describes companywide hackathons and training that enable PMs and even Customer Success to ship production changes using agentic engineering tools (e.g., Devin).

    • Leadership sets expectation: building is for every function
    • Companywide hackathons as repeated ritual (quarterly/6 weeks)
    • Training: how to ship to production safely even if non-technical
    • Examples: PMs shipping front-end + back-end features; CS contributing via guides
    • Break work into playbooks, then into skills/agents for repeatability
  9. PMs shipping real features: ontology of PM work and what gets automated away

    JZ shares how Laurel redefines PM work to look more like engineering: feature work, testing, QA, backlog execution—augmented by agents. Meanwhile, historically PM-heavy tasks (competitive analysis, stakeholder writing, research ops) should be automated or delegated to workflows.

    • PM ontology intentionally mirrors engineering responsibilities
    • PMs own end-to-end quality: build, test, QA, fix bugs
    • Automate: competitive intel gathering, synthesis, scheduling, repetitive stakeholder comms
    • ‘Build the system once, monitor it’ vs. doing manual deep work daily
    • OS surfaces skills just-in-time to change behavior
  10. The Captain Model: assigning ownership by the most critical skill for success

    JZ introduces the ‘captain’ concept: each initiative has a single accountable leader chosen based on the hardest part to get right (architecture, interaction design, content/business context). AI tools help non-engineers assess risk and pull in engineers for code review where needed.

    • Captain chosen by what matters most: engineering, design, data, PM judgment
    • AI tools can analyze codebase state and highlight risks
    • Non-engineers can ship when back-end is stable and risk is understood
    • Engineers still review risky parts; governance is integrated, not removed
    • Captain model reduces handoffs and accelerates delivery
  11. Governance without bureaucracy: transparency, guardrails, and two-track product reviews

    To prevent conflicts and quality issues, Laurel uses transparency (shared channels, visible PRs) and lightweight rules. JZ explains a two-track review system: small changes move fast with async checks; strategy/architecture shifts get rigorous reviews to preserve global coherence.

    • Transparency: channels for Devin/PR visibility and reviewer tagging
    • Guardrails: enablement guides + quick pre-checks for proposed changes
    • Two-track model: fast lane for small features; heavy review for systemic changes
    • Compress product reviews into async feedback loops when possible
    • Reject ‘no roadmap/no planning’ extremes—strategy still matters
  12. Strategy example: expanding the meaning of time + operationalizing ‘unreasonable hospitality’

    JZ describes product strategy reviews focused on where Laurel wins and expands—time tracking beyond billable hours to broader ‘time as a resource’ across roles. She also explains how the OS can systematize culture, like ‘unreasonable hospitality,’ by prompting personalized customer delight actions using captured context.

    • Strategic expansion: time matters even outside billable-hour firms (and in tech)
    • Humans focus on high-value work (relationships, onsite hospitality)
    • Automate logistics and prep to enable better human moments
    • Operationalize values: ‘unreasonable hospitality’ becomes a repeatable practice
    • Use transcripts/context (e.g., Gong) to suggest tailored delight ideas
  13. Transformation playbook for non–AI-native companies: start small, build allies, scale 1% workflows

    JZ argues every company will be forced toward AI-native velocity, but individuals can move first without CEO-level access. She recommends starting with small automations, building tools for other teams, rapidly generating playbooks with LLMs, and celebrating power-user workflows to spread adoption.

    • Competitive pressure will force change; don’t wait to be dragged
    • If you can’t ship code yet, build tools/workflows for another hungry team
    • Use LLMs to draft playbooks fast; refine with business-specific judgment
    • Create a culture of celebrating automation wins and sharing workflows
    • Scale the ‘1%’ experiments into standard operating practice
  14. AI Ops as the new Biz Ops: the ‘Sasha model’ and CEO-level conviction

    JZ explains how Laurel institutionalized AI adoption by creating an AI Operations function—full-time people whose mandate is workflow automation and tool-driven efficiency. She credits the CEO’s early conviction to re-architect the company around AI, while noting leaders at any level can still drive function-level transformation.

    • AI Ops: dedicated role to tinker, build workflows, and prove value
    • ‘Sasha effect’: one great AI operator creates demand across every function
    • Path to scaling: AI Ops per domain (GTM, product, finance, etc.)
    • CEO conviction accelerates org-wide shift; Laurel re-architected post-2018 founding
    • Functional leaders can still map work, color-code automation opportunities, and drive change
  15. Hiring the future PM: small orgs, senior builders, screen-share interviews, and 4 AI maturity levels

    JZ forecasts leaner product orgs with more senior, judgment-rich builders who love hands-on work and can ship end-to-end. She details her screen-share interview to verify real AI practice and outlines four levels of AI maturity from chat usage to shared apps and customer-shipped workflows.

    • Trend: smaller orgs reduce coordination costs; more end-to-end ownership
    • Senior ‘ICPM’ builders can outproduce larger teams when AI-enabled
    • What she looks for: battle-tested judgment + curiosity + fearlessness + customer closeness
    • Interview tactic: screen-share to reveal real usage vs. buzzwords
    • Four AI levels: (1) chat/search, (2) workflow automation, (3) app building, (4) shared apps/shipping to customers

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.