Show Notes
This daily breakdown shows how to write PRDs like the Airbnb founder: start with the problem, push for an audacious, AI-first product, and use 11-star experience plus pirate metrics to guide product thinking.
Problem-first PRD mindset
- Start with the real problems you’re solving for your community or users.
- Translate pain points into concrete, testable requirements.
- Example focus from the video:
- Content management is outdated and hard to navigate
- Discovery and onboarding flow are painful
- Lack of centralized prompts and data ownership
- Limited analytics and no robust API/AI-first capabilities
- Outcome: a clear, AI-first product concept that can scale to web and iOS (Android later)
Airbnb 11-star experience (planning technique)
- Concept: sketch the best possible version of the product experience, and push it up to 11 stars.
- Purpose: stretch thinking beyond “good enough” and uncover hidden UX levers.
- Why it matters: helps prioritize what to build first and what would truly differentiate the product.
Agentic workflows and pirate metrics (AARRR)
- Agentic workflows: design automated, goal-driven processes for each funnel stage (acquisition, activation, retention, revenue, referrals).
- Pirate metrics (AARRR) to shape the product:
- Acquisition: how do users find you? growth agents, cross-platform touchpoints
- Activation: first meaningful moment (FTX) and the aha moment
- Retention: repeat use and ongoing value
- Revenue: membership tiers, monetization strategy
- Referrals: how users bring others in
Acquisition ideas (agentic growth)
- Growth agents across channels: FAQs answered, cross-posts on YouTube comments, Reddit, forums
- Outbound vs platform-owned channels: capture the lead, tie to email for nurture
- Programmatic SEO and content loops: track what people actually search for and answer it automatically
Activation and the aha moment
- Define the first-time experience (FTX) and the moment users feel real value.
- Examples: following a threshold (e.g., 13 follows) or reaching a specific milestone that signals value.
- Tie onboarding to a quick win: users should see tangible value fast to reduce drop-off.
Retention, revenue, and referrals
- Retention: track how long users stay engaged and what keeps them coming back
- Revenue: design tiered offers that scale with value and usage
- Referrals: incentives or social proof mechanisms that encourage sharing
- The key: use data to understand what drives ongoing value, not just traffic
Design and tooling for an AI-first platform
- Screens and flows: start with sketches (Figma is still useful)
- Prompts store concept: store prompts as from-to pairs, with model selection and versioning
- From → To: track inputs and expected outputs
- Model selection: choose the right model per task
- Outputs side-by-side: compare results across models
- Readability and messaging: a Hemingway-style editor to grade readability and highlight improvement opportunities
- Target: communicate at a lower grade level for broad audiences
- Data and analytics: user data ownership, vector search, personalized feeds, and event tracking
- Onboarding workflows: agentic steps for education, content discovery, and value realization
Practical steps to implement
- Start with a concrete PRD problem statement and a short list of top 3-5 solutions
- Apply the 11-star exercise to push the UX beyond the obvious
- Map out pirate metrics for your funnel and design agentic workflows for each stage
- Draft the initial screens and a minimal prototype to test the core idea
- Build a prompts ecosystem: store, versioning, and from-to prompts with model options
- Add a Hemingway-style readability check to speed iteration
Next steps and mindset
- Don’t chase shiny tech too early—clarify what you’re solving and why it’s valuable
- Use the 11-star exercise to unlock ambitious UX without overengineering
- Use PRD + agentic workflows to stay aligned as you prototype and build
Call to action
- If you’re excited about this AI-first platform concept, drop your thoughts in the comments.
- If you want longer-form walkthroughs, say “make more long-form” or drop a rubber duck to signal you want depth.