Solving the problem of "Authentication" for Custom GPTs
Solving the problem of "Authentication" for Custom GPTs
We are using PostHog, a product analytics tool, as a use-case to solve the problem of Authentication for Custom GPTs.
PosthogGPT can watch session recordings.
PosthogGPT can watch session recordings.
Session recordings are one of the best ways to discover product problems, but watching those videos takes hours, especially when you have a sizeable user base.
We have built a custom GPT for PostHog which can watch session recordings. It tells exactly how long the user was active/inactive, technical errors and what needs to be improved in the product. Pre-GPT era this was not possible.
We have built a custom GPT for PostHog which can watch session recordings for you. It tells exactly how long the user was active/inactive, technical errors and what needs to be improved in the product. Pre-GPT era this was not possible.
Why Authentication?
Why Authentication?
Why Authentication?
To share PostHogGPT on GPT store an Authentication layer is needed. PostHogGPT calls the PostHog API to retrieve session recordings. User of PostHogGPT should be able to connect to their own PostHog account. A GPT builder will have to either build authentication (OAuth) by themselves or they could use Teko. In short, distribution in such use case is impossible without end-user Authentication.
