Move from Pinecone to Axonio.
Every record ID in the index, then the stored vectors and metadata in batches. Map it once, review a sample, and start. Your agents keep everything they knew, and gain memory they can branch.
You’ll need
- A Pinecone API key
- The index host URL
- Optionally, a namespace
Steps
Import from Pinecone.
- 01Open Imports and choose Pinecone.
- 02Paste an API key and your index host (the URL shown on the index page in Pinecone).
- 03Pick the metadata field that holds the text. Axonio guesses it from a sample; you confirm.
- 04Start the import. Existing embeddings are kept when their dimensions match your Axonio database.
What you get that Pinecone 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 Pinecone 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.