He raised over $50M for TeamFlow, then fired two-thirds of his team when COVID ended. Flo Crivello pivoted to building AI products with Lindy, an agent platform that lets anyone automate workflows without code. The first version was so broken it sent emails saying "the user wants me to send an email to 50 software engineers."
Flo reveals the AI product development lessons that took Lindy from a broken V1 to high 7-figure ARR, including the "Notion head fake" positioning strategy that made building AI products accessible by positioning against something familiar. You will learn why LLM products need to start with familiar positioning, how shipping embarrassingly broken AI features helped find pioneers, and when to rebuild everything at 100K MRR.
Flo previously worked at Uber and spent years building a Twitter audience with 20 tweets a day. A single demo video converted that audience into 70,000 waitlist signups for Lindy in March 2023.
š Key Lessons
- šÆ Position your AI product against the familiar, not the alien: "AI employee" was too futuristic for a broken product. "If Zapier and ChatGPT had a baby" tapped into existing mental models when building AI products.
- š Ship an embarrassingly broken AI product to find pioneers: Lindy V1 sent emails that literally quoted the user's instructions. Early adopters forgave it because they bought the vision.
- šŖ Climb the ladder of abstraction when building AI products: Lindy started as "update Salesforce after meetings," then generalized to any CRM, then any tool. Start specific, then keep generalizing.
- š± Build audience before launch with daily social content: Flo tweeted 20 times a day using a script to track volume. One demo video converted years of audience into 70,000 waitlist signups.
- š° Rebuild at 100K MRR if the AI product paradigm is broken: Flo spent 5-6 months rebuilding because the architecture could not deliver. When 99.9% of revenue is in the future, do not optimize for the present.
Chapters
- Introduction and what Lindy does
- Flo's background at Uber and TeamFlow
- How COVID killed TeamFlow's growth
- How the idea for building AI products emerged from Salesforce automation
- The brutal pivot: firing two-thirds of the team
- Launching with a demo video and 70,000 waitlist signups
- When the AI product did not work
- Positioning: from AI employee to Zapier of AI
- The Notion head fake strategy
- Rebuilding while serving customers
- When MattVidPro's YouTube video accelerated growth
- Lightning round
Resources
