$150K ARR. Customers never logged in. They'd call with a question, get an answer, and disappear. Ibby Syed spent 18 months building what he thought was an AI SaaS - then realized he'd accidentally built a consulting business. The wake-up call came when 100 lines of OpenAI code replaced his entire data science solution.
Ibby reveals the exact moment that triggered the AI SaaS pivot, why teaching customers to build their own AI agents scales better than building for them, and the outbound strategy where he sends actual leads from Reddit monitoring before the first call.
Cotera is an AI-powered platform that lets enterprise customers build prompt-based AI agents on top of their existing data warehouses. The AI startup has 15 enterprise customers and generates over $1M ARR with a team of 10.
This episode is brought to you by:
💖 Gearheart → Book a free consult and get the first 20 hours free
🚨 NordStellar → Book a demo and get 20% off with code blackfriday20
🔑 Key Lessons
- 🚨 Recognize when your AI SaaS is actually consulting: Ibby hit $150K ARR but customers weren't logging in. They called for answers instead of using the product - a dangerous signal he almost ignored.
- 💡 Let API breakthroughs trigger your pivot: Ibby's co-founder solved a customer problem with 100 lines of OpenAI code that outperformed a complex data science solution. That contrast made the AI startup opportunity obvious.
- 🎯 Deliver value upfront in outbound: Instead of pitching, Ibby sends actual leads from a Reddit monitoring AI agent. Showing value before the first call converts better than any cold pitch.
- 🛠️ Teach customers to build, don't build for them: After the pivot, Cotera stopped doing custom implementations. Teaching customers to build their own AI agents is what made the AI SaaS business scale.
- 🏢 Enterprise customers want AI on their own infrastructure: Series B+ companies want AI-powered platform capabilities on their existing Snowflake or BigQuery, not third-party clouds.
Chapters
- Introduction and the "White Collar" quote
- The Y Combinator journey and the first idea
- Getting first customers through LinkedIn outbound
- The consulting trap - revenue vs. scalability
- The wake-up call - 100 lines of code vs. data science
- The pivot to building an AI SaaS agent platform
- The "teach, don't do" service model
- Prompt-based workflows vs. drag-and-drop
- Why vertical AI startups might die
- Making AI agents work at scale
- Lightning round
Resources
