With Open WebUI, each person can bring their old chats across: export your data from ChatGPT, then in Open WebUI open Settings, Data Controls, Import Chats and choose the conversations.json file from the export. Chats come across; custom GPTs, memory and uploaded files do not, so rebuild the custom GPTs your team uses as custom models with the same instructions and knowledge files. Before inviting anyone, the admin adds an AI connection under Admin Panel, Settings, Connections, for example an OpenAI API key or any OpenAI-compatible service. Then add your colleagues, or turn on sign-in through your identity provider, and put them in groups for the assistants they share.
LibreChat imports chats too: each person exports their data from ChatGPT, then opens Settings, Data and Privacy, Import and chooses the conversations.json file. Custom GPTs, memory and files do not come across, so rebuild the custom GPTs your team relies on as agents and share them with the right users or groups. The first account you create becomes the admin; connect at least one AI provider with its API key before inviting colleagues.
AnythingLLM cannot import ChatGPT chats. In its first-run setup, choose your AI provider and paste its API key, then create a workspace for each set of documents your team asks about, upload the files, and add colleagues to the workspaces they need.
Onyx cannot import ChatGPT chats either. The first person to sign up becomes the admin: connect an AI provider with your API key, turn on invite-only sign-up, then add connectors for the apps your team asks about, such as Slack or Google Drive, giving Onyx access in each. Large sources take a while to index the first time.
LobeHub cannot import ChatGPT chats. Each person creates an account and adds an API key for a provider such as OpenAI under Settings, Provider, then rebuilds the custom GPTs they use as agents with the same instructions, and uploads their files into a knowledge base.
HuggingChat cannot import ChatGPT chats. When you install it with Cloudgate, choose the model provider and paste its API key, such as a Hugging Face access token, and its models appear on the Models page. Your browser then asks for a login: the username is admin and the password is the one Cloudgate generates, and everyone you share it with uses that same login. For a sign-in per person, set up OpenID Connect with your identity provider and limit it to your company's email addresses or domain.
Khoj cannot import ChatGPT chats. Sign in to its admin page, add an AI provider key and a chat model, then upload your files or sync them from the desktop app, Obsidian or Emacs. Rebuild the custom GPTs you rely on as Khoj agents with the same instructions and knowledge files.
Synaplan cannot import ChatGPT chats. As admin, paste an API key from a provider such as OpenAI, Anthropic or Groq under its AI settings, upload the files your team asks about, and rebuild the custom GPTs you use as assistants with the same instructions and knowledge folders. Then add your colleagues and share folders and assistants with their groups.
Eneo cannot import ChatGPT chats. Sign in with the admin email you gave during setup and the password Cloudgate created for it, then add your AI provider and the models staff may use under Admin, Models. Turn on sign-in through your identity provider, create a space for each team with the documents or SharePoint libraries it works from, and rebuild the custom GPTs your teams use as assistants in those spaces.