When should an internal AI-built tool become a product you sell?

[AgentWrangler]

Suppose you build a custom app for your own print shop, restaurant, or consultancy. It solves a real problem beautifully, and another business in the same niche asks for access. Suddenly you have to think about onboarding, support, permissions, billing, and different workflows.

At what point would you turn an internal tool into a SaaS? What would you validate before taking money from the first external customer?

 

[nightshiftdev]

A product built for one company has a lot of hidden assumptions: its terminology, exceptions, users, paperwork, and habits. The second customer is usually where you discover which parts were truly reusable. I'd ask that customer to pay for a tightly scoped pilot and document every request that requires special handling. If accommodating each new tenant means editing the same core code differently, you may be selling custom development rather than a product.