AI search & your indexing key
NextELN search has two modes:
- Keyword search — always on, free, needs no key. Matches the words you type.
- Semantic (AI) search — understands meaning, so "bacterial expression host" finds your BL21/DH5α clones even if those exact words aren't written down. This needs your notebook to be embedded into a vector database.
Why an indexing key — and who pays
Embedding your data is a paid AI operation (billed by the AI provider per token). To keep this fair and transparent, each workspace uses its own key and pays that provider directly. It's tiny: indexing a full notebook is typically a few cents, and a search is a fraction of a cent.
Choose your provider — OpenAI or Voyage
Go to Admin → AI → Indexing key and pick whichever you prefer:
| Provider | Get a key | Notes |
|---|---|---|
| OpenAI | platform.openai.com → API keys | Add a little credit under Billing (even $5 lasts a long time). If you already use an OpenAI key for chat, it can double as your indexing key. |
| Voyage | dashboard.voyageai.com | Free tier: 200M tokens, no credit card — usually enough that you never pay. |
Paste the key and click Save indexing key. We run a 1-token test embed before saving, so a key with no billing/quota is rejected right away with a clear message (no silent failures later).
What's free vs. paid
Keyword search is always free. Semantic (AI) search runs on paid embeddings, so it uses your own indexing key — you pay your provider (OpenAI or Voyage) directly, at wholesale, typically pennies. That keeps you fully in control of the cost, with no middleman. Better yet, Voyage's free tier (200M tokens, no card) means most labs pay nothing in practice — and the key is always yours.
Turn it on
- Admin → AI → add your OpenAI or Voyage indexing key (or confirm the platform covers you).
- Make sure Enable vector database is checked.
- Click Rebuild vector database once. You'll see "…with vector embeddings — semantic search on."
- Go to Search and try a natural-language query.
If you ever see "keyword only" after rebuilding, your key is out of quota — top it up and rebuild again.
Need help? support@nexteln.com — or ask the in-app AI Lab Assistant.