How Perplexity builds product

Lenny Rachitsky interviews Johnny Ho, co-founder and head of product at Perplexity, about how a fast-growing AI search company builds product with a very small team. Perplexity uses AI to answer its own operational questions, keeps teams to two or three people, relies on self-driven individual contributors rather than managers, and organizes work to minimize coordination costs. Planning is quarterly with weekly 75% goals, decisions are decentralized around a single directly responsible individual, and feedback is shared asynchronously. The approach matters because it previews how many companies may structure product work as AI reduces the need for coordination overhead and elevates technically skilled product generalists.

01Key takeaways

• Use AI to answer routine operational questions before interrupting colleagues, which saves time and reduces coordination load.
• Break projects into self-contained parallel tasks so people can execute without waiting on others.
• Set weekly goals aiming for 75% completion, and treat the remaining gap as a signal about prioritization or staffing.
• Hire self-driven individual contributors with measurable user impact rather than people strongest at managing processes.
• Keep decisions decentralized with a single owner per project, and resolve misalignments quickly afterward instead of seeking consensus upfront.
• Document reasoning in writing to make decisions clearer and reduce the need for meetings.

02Key sections

AI as an operating assistant

Perplexity used AI from the start to learn unfamiliar functions like launching products and running HR, though its coding help was limited to templates. Employees now ask AI before interrupting colleagues.

Small teams and few PMs

Projects typically involve one to three people, with only two full-time PMs in a company of about 50. PMs are most valuable for hard branching decisions about AI use cases.

Hiring for initiative and IC strength

The company values flexibility and quantifiable user impact over coordination or process skills, and expects technical PMs and engineers with taste to become most valuable.

Slime-mold organization and coordination costs

Teams are structured to minimize coordination headwind by parallelizing work, sharing reusable processes, and trusting people to reach out directly. Plans are quarterly and roadmaps stay flexible given fast AI change.

Decentralized decisions and tooling

Each project has a single DRI, work is broken into self-contained parallel tasks in Linear, and iteration happens asynchronously in Slack before launch via dogfooding. Notion holds docs and postmortems.

03From the post

“1. The secret to Duolingo’s exponential growth 2. How to accelerate growth by focusing on the features you already have 3. How AI will impact product management 4. How to make an impact in your first 90 days For $150 a year, get access to these posts and every prior post, along with an invite to a private Slack community with global meetups, a mentor matching program, interview prep support, live AMAs, and more. I guarantee you’ll get 100x the value of a subscription or your money back. Founded less than two years ago, Perplexity has become a many-times-a-day-use product for me, replacing many of my Google searches—and I’m not alone. With fewer than 50 employees, the company has a user base that’s grown to tens of millions. They’re also generating over $20 million ARR and taking on both Google and OpenAI in the battle for the future of search. Their recent fundraise of $63m values the company at more than $1 billion, and their investors include Nvidia, Jeff Bezos, Andrej Karpathy, Garry Tan, Dylan Field, Elad Gil, Nat Friedman,…”

04Frameworks mentioned

Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.