Skip to content

Move from Chroma to Axonio.

IDs, documents, embeddings and metadata from the collection. Map it once, review a sample, and start. Your agents keep everything they knew, and gain memory they can branch.

You’ll need
  • Your Chroma server URL
  • The collection name

Targets self-hosted Chroma. For Chroma Cloud, export to JSON and use the file importer.

Steps

Import from Chroma.

  1. 01Open Imports and choose Chroma.
  2. 02Enter the server URL and collection name.
  3. 03Review the mapping and start.

What you get that Chroma doesn’t give you.

Version what agents know

Fork memory to test a policy, diff it against main, then merge it or throw it away.

Relations, not just vectors

Imported content is linked into a knowledge graph as it lands, so recall can follow entities and time.

MIT core, run it anywhere

Same engine in our cloud, your VPC or on-device. No lock-in on the way out either.
FAQ

Migration questions.

Do I lose my Chroma embeddings?

No. Existing vectors are kept when their dimensions match your Axonio database; otherwise content is re-embedded on import so search quality is consistent.

Is anything written before I confirm?

No. You see a live sample of what will be imported and the field mapping first. Nothing is written until you start the job.

What happens to my credentials?

Keys and connection strings are encrypted at rest and used only by the import job. The file importer needs no credentials at all.

Can I run both side by side while I switch?

Yes. Import a snapshot, point one agent at Axonio, and compare. Re-running an import is safe; it reports what was imported and what was skipped.

Your Chroma memories, branchable.