As a (paying) ChatGPT user I was surprised to find that there were no convenient way to add an LLM-powered chatbot (GPT or otherwise) to Slack (chat platform I use most often), so I wrote one.
Geppetto currently uses gpt-3.5-turbo model via OpenAI API and answers when you mention it (@geppetto) with a question. Each conversation thread is like a saved conversation within ChatGPT interface.
As a privacy-sensitive European, I tried to make this as least data-intrusive as possible: Gepetto stores nothing (I literally have no idea who uses it, I just get opaque IDs from Slack, and I only get your email if you subscribe, for payment/subscriptions purposes), and it only listens to conversations where it was explicitly mentioned (or DM'd).
I have a big todo list (more models, gpt4, multi-model features, plugins) but as a MVP, this has already been very useful to me (I built Geppetto using Geppetto to help me out with stuff like Slack integration, Next.js setup, etc), and I hope it'll be useful to others.
Hey HN!
As a (paying) ChatGPT user I was surprised to find that there were no convenient way to add an LLM-powered chatbot (GPT or otherwise) to Slack (chat platform I use most often), so I wrote one.
Geppetto currently uses gpt-3.5-turbo model via OpenAI API and answers when you mention it (@geppetto) with a question. Each conversation thread is like a saved conversation within ChatGPT interface.
As a privacy-sensitive European, I tried to make this as least data-intrusive as possible: Gepetto stores nothing (I literally have no idea who uses it, I just get opaque IDs from Slack, and I only get your email if you subscribe, for payment/subscriptions purposes), and it only listens to conversations where it was explicitly mentioned (or DM'd).
I have a big todo list (more models, gpt4, multi-model features, plugins) but as a MVP, this has already been very useful to me (I built Geppetto using Geppetto to help me out with stuff like Slack integration, Next.js setup, etc), and I hope it'll be useful to others.