ChatGPT charges for access. Its subscription plans range from $8 per month for Go to $20 for Plus and $100 or more for Pro, with plan-dependent usage limits. Customers pay a predictable monthly fee regardless of whether the AI produces useful or accurate results, bearing the risk of paying for a service that may not meet their needs. OpenAI, meanwhile, covers infrastructure costs and manages its risk through usage limits and additional usage options.
GitHub Copilot charges per person. Its Business and Enterprise plans cost $19 and $39 per user per month, respectively, with AI credits included. Companies pay based on the number of users rather than access to a shared service or individual outcomes. As with ChatGPT, customers bear the risk of inaccurate or unhelpful results, while GitHub covers service costs and manages compute expenses through usage limits and additional credits.
Intercom Fin charges for outcomes. Customers pay $0.99 per resolved conversation, meaning they pay based on whether AI successfully resolves an issue rather than simply providing access. This shifts more value risk to Intercom since unresolved conversations generally aren’t charged. However, as Kyle Poyar points out, disagreements may arise over what qualifies as resolved, leaving both sides to negotiate how success is measured.
For our consensus tool, a hybrid pricing model makes sense. I would charge a platform fee per team that includes a set number of core members. Occasional participants, such as executives reviewing decisions, could receive free, limited-access seats. This draws on Poyar’s ideas about lite users and minimum-user commitments.
AI analysis would draw from a shared pool of team credits, with additional credits available during busier months. I would avoid charging per resolved conflict because defining resolution could be subjective and lead to disagreements.
