interface0 continues to be my preferred way to interact with LLMs (and increasingly a few of my friends’ preference as well!). Here’s some of what I shipped since the previous post:

Chat sharing

You can now share chats, either with another individual interface0 user, or a team (see below). When you share it, you can choose whether they are a “viewer” or an “editor.” Either way, the chat will now show up in their interface0 inbox, and if they make them an editor, they will also be able to send messages in the chat.

Sharing makes it easy to have collaborative conversations with AI, with multiple users in the chat. Plus, anyone can “fork” the chat at any time to split it off into their own private conversation.

This alone feels like a huge upgrade over existing LLM interfaces for certain use cases (especially within teams).

You can also publish chats publicly and share a link with anyone, including non-interface0 users.

Knowledge

You can now create a folder with text and files that gives the agent knowledge, and tag that in on any specific chat request. This is similar to ChatGPT/Claude’s “projects” feature, except that the chat doesn’t need to be in a project folder, and you can tag multiple pieces of knowledge into any given chat.

I have a “health” knowledge folder with info about my health, plus ones for different companies, and it makes it easy to tag that context into any chat where I need it.

Teams functionality

You can now make “teams” — for example, we now have an interface0 team at Amity Ventures, where I’m a venture partner, and at Lightwork. You can be on multiple teams.

Once you’re on a team with other interface0 users, a few options open up. You can share contexts/knowledge folders, template messages, and personas across the team, so everyone can use them.

For example, we can have a “water testing protocol” knowledge folder at Lightwork, or a “memo drafting” template message at Amity. Everyone on the team could access these.

Plus, chats can be shared with a whole team, as viewers or editors. That way the chat shows up in each person’s inbox on the team.

Lots of little features

Some favorites:

  • Enhance prompt — just hit a button and your half-assed prompt gets upgraded into something more clear for the agent
  • Autopilot mode — you can let an agent select what model to use, what knowledge to loop in, and what persona to leverage for each message instead of choosing yourself
  • Summarize chat — hit one button and get a summary of the current chat, which can then be used to kick off a new chat or to create a new knowledge folder. You can also download a chat as a text file for use elsewhere.
  • Template variables — template messages can now have variables in them. When you select the template message, you just fill in the variables in a popup and the completed message gets inserted
  • Forwarding email — every user has an email address they can send or forward emails to. When they do, it kicks off a new chat where the agent suggests next steps on the email or answers whatever question you included
  • Edit message — you can edit the last message you sent in the conversation, which will cause the agent to update their reply
  • A better landing page — check it out!

Plus a whole lot of bugfixes, performance improvements, and other quality-of-life touches…

It’s great fun building “personal software” like this to scratch my own itches. I can create exactly the interface I want to use. It will continue to evolve! Ping me if you want an invite.

Looking for more to read?

Want to hear about new essays? Subscribe to my roughly-monthly newsletter recapping my recent writing and things I'm enjoying:

And I'd love to hear from you directly: andy@andybromberg.com