ChatGPT, GitHub Copilot, and Intercom Fin all sell AI, but each one charges for something different. ChatGPT charges for access: you pay a flat monthly fee for a tier, and the price stays the same no matter how much you use it. OpenAI absorbs the cost of heavy users, which is why higher tiers mostly buy higher limits, but if an answer is wrong, the user has still paid. Copilot charges per person, at $10 to $100 per user per month. GitHub is betting each developer is worth more with it. Each plan includes a fixed amount of AI credits, so GitHub carries the cost risk up to that cap and the buyer pays past it. Bad code is still the developer’s problem. Fin charges for outcomes, at $0.99 each, such as a conversation where the customer needs no more help after Fin’s last answer. If there’s no outcome, there’s no charge, so Intercom carries most of the risk. The catch is that a customer who gives up and leaves can look like a resolution. For our team’s product, a platform where AI sends each piece of feedback to the right reviewer, per-seat pricing fits best. Its value grows as more of a team uses it, and outcome pricing would start fights over what counts, since “feedback reached the right person” is hard to measure. To cover heavy AI use, we’d include a routing allowance per seat, like Copilot’s credits.
