Skip to content

Memory

One memory for every agent you use

Gryz memory is one place for the decisions and preferences you would otherwise repeat in every chat. A rule Claude Code saves on Monday is there when Cursor or ChatGPT needs it on Tuesday, and you can see which agent saved each memory, where it came from, and change or forget it.

Memory · Project: web-app

Deploys go from develop, never main.

Saved by Claude CodeFrom Launch plan v3

“Deploy from develop after the checks pass. Never from main.”

Release notes go out at 9am on launch day.

ChatGPTPersonal
Example: a saved memory, "Deploys go from develop, never main.", marked as saved by Claude Code with a link to the doc it came from, and an older memory below it.

Memory

What you can do

Saved by one agent, used by all
Agents save with remember and look things up with recall. Every agent you connect reads and writes the same memories, so you stop pasting the same background into each new chat.
Personal or per project
Personal memories follow you into every tool. Project memories stay with one repo or workspace, so a rule for one app stays with that app.
See where each one came from
Every memory shows who saved it, you or one of your agents. When an agent learned something from one of your docs, it can link that doc and quote the passage, and Gryz flags the memory if the doc changes later.
Yours to change
Edit a memory, export all of them, or forget one for good. Forgetting removes the text and its earlier versions instead of just hiding them.
Kept private
Your memories stay in your account and nobody else sees them. Gryz never uses them to train models.

Claude, ChatGPT, Cursor and Codex

How it works with your agents

Connect Claude, ChatGPT, Cursor or Codex once, and the agent checks memory before it asks you to repeat yourself. In Claude Code or Cursor, run Gryz setup inside your repo so project memories line up with that repo, and every tool sees the same facts about the same code.

Agents are told to ask you before they forget anything. Whatever comes back from memory is treated as information to weigh, never as an instruction to follow, so a stray note in a memory cannot steer your agent.

Memory

Memory for a whole team

Company workspaces, which we are opening to teams one at a time, keep a shared memory of their own that follows the company's permissions. If the doc a company memory came from changes or is deleted, the memory lands under To check in the review inbox for the people who can edit that project, so a stale rule gets caught before an agent leans on it.

Questions

Does this replace the memory inside ChatGPT or Claude?

No, it sits beside it. The memory inside one app stays in that app. Gryz memory is the part every tool you connect can read, which makes it the right home for rules and decisions your whole setup should follow.

What happens when the doc a memory came from changes?

The memory is flagged so agents know its source moved. Before relying on it, they are told to read the doc again, then either save the memory fresh or ask you whether it should go.

Can I use memory on a free account?

Yes. Any Gryz account can save and recall memories from a connected agent. Paid plans raise how much you can keep, and the pricing page lists the limits.

Connect your agentsSee every feature

Free with a Gryz account. About 2 minutes.