One thing I found interesting about these three products is how differently they charge for essentially the same thing. They are all selling access to AI and have different approaches. With ChatGPT, you’re paying a fixed monthly subscription, which is nice because you know what to expect. But that also means OpenAI has to deal with the cost of people using it a lot more than others. GitHub Copilot charges per developer instead, which makes sense for companies buying it for their teams, although they’re still paying for developers who might barely use it or find the suggestions unhelpful.
Intercom Fin’s model stood out to me the most because it charges for each customer conversation that gets resolved. Instead of paying just to use the AI, companies are paying for something it actually accomplishes. I think this puts more risk on Intercom because if the AI can’t resolve an issue, they don’t get paid for it, even though they’ve spent money running the model.
The reading also brought up something I hadn’t really considered, which is how difficult outcome-based pricing can be in practice. It sounds great to only charge customers when they get value, but figuring out what counts as a successful outcome isn’t always straightforward.
For Chilla, I think a per-seat model would probably make the most sense, where companies pay based on how many employees have access to it. Since we’re helping people find the right colleagues across teams, I don’t think charging per introduction or meeting would work very well. It might actually discourage people from using Chilla, and it’s hard to measure whether a connection was genuinely useful. One concern with per-seat pricing, though, is that companies might hesitate to pay for employees who don’t use it regularly.
