← Back to Blog
Article
September 15, 2026

If you use more than one AI assistant (like ChatGPT and Hermes, or Claude and ChatGPT), you've likely run into a problem: you tell one of them something important, and the others never find out.

How to share context between AI agents.

Maybe you use Claude Code for one project, Cursor for another, and a third tool for something else entirely. Each one keeps its own notes about you. None of them talk to each other. Here's a great example from one of our employees. So when your deadline moves, or a meeting gets rescheduled, or a project gets put on hold, you're the

There's a fix that's gotten popular lately: a shared file called AGENTS.md that different AI tools can all read. It's an  improvement. But it only solves half the problem.

AGENTS.md is only a file.

It holds whatever you wrote down, whenever you last wrote it, but it cannot update in real time. As a result it can go stale. Now, all your tools are all wrong together instead of wrong separately.

We've solved this exact problem before — just not for AI

Back in the early days of PC software, every program had to know how to talk to every printer. WordPerfect, for example, ended up maintaining huge lists of printer-specific drivers just so your document would print correctly. That got fixed by building printer support into the operating system itself, so no individual program had to worry about it anymore.

Some people in the industry think AI tools are stuck at that same stage today. Michael Tiffany, who co-founded Fulcra Dynamics, has been working to fix this: instead of every AI tool keeping its own guesses about your life, there should be one place underneath all of them that holds what's  current — timestamped, updated, and available to whichever tool needs it.

What that tool actually looks like

Fulcra and a few similar companies now offer this as a service. Instead of editing one shared file, you (or one of your AI tools) add a new note whenever something changes. Any other connected tool can then check what's changed since it last looked, instead of re-reading everything from scratch and hoping it's still accurate.

How to try it yourself

  1. Connect one AI tool to a shared context service — Fulcra's hosted version lives at https://mcp.fulcradynamics.com, or you can run it yourself with one command (uvx fulcra-context-mcp@latest) if you'd rather keep your data local.
  2. Tell that tool something new — a fact, a decision, a change of plan.
  3. Open a second AI tool, connect it to the same service, and ask it about that same thing — without telling it yourself.
  4. If it gets the answer right, the two tools are actually sharing a picture of your life instead of each guessing on their own.

The takeaway

This is the future for agent sharing; get onboard.

The future is personal and private.

Fulcra was designed by people who get privacy and know the importance of an infrastructure solution that can be the secure private datastore for the rest of your life. Here data is yours, under your control, and only shared with the people and tools you choose to share it with.