Free Intro Class|BotSpot Builder Program|Learn. Build. Earn.|Master AI Agents • Build Real-World Projects • Industry-Ready Portfolio Limited Seats Available
agenthub · knowledge.embed
Knowledge Embed
Attach an embedding vector to every chunk. This is the step that costs model money in the ingestion chain.
category Knowledgein chunksout chunksexecution asyncmodel cost model tokens
Why the node earns its place
Chunks arrive without vectors and leave with them. The vector width is the adapter's business — the node never asserts one — so the same graph runs against different embedding backends unchanged.
How it works at run time
- 1Embed each chunk's text,
concurrencyrequests at a time. - 2Drop any chunk whose embedding call failed, and log it.
- 3Emit the chunks with vectors attached.
What you wire
Inputs
- chunks *array
Outputs
- chunksarray
Configuration
| Key | Type | Default & options | What it does |
|---|---|---|---|
| concurrency | number | default 4 | Concurrent embedding requests per batch. |
What usually goes wrong
Watch for this
Failures drop chunks rather than failing the run — partial success beats total failure for an indexer, but it means a quiet gap in your knowledge base. Compare the chunk count in and out before treating an index as complete.
Behaviour & provenance
buildersChat & voice agents
routesno
side effectsnone — computes only
replay safetynot applicable
talks toa model
holdsworkspace runtime token