Skip to main content
POST
/
brain
/
ingest
curl --request POST \
  --url https://api.atlas.bsyncs.com/brain/ingest \
  --header 'X-API-Key: atlas_your_key_here' \
  --header 'Content-Type: application/json' \
  --data '{
    "text": "Project Apollo uses PostgreSQL on AWS RDS.",
    "user_id": "user-123"
  }'

{
  "facts_ingested": 1,
  "episodic_chunks": 1,
  "entities_extracted": 2,
  "triples_extracted": 1,
  "latency_ms": 980
}

​
Authorizations

​
X-API-Key
string
required
Your Atlas API key. Format: atlas_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx.
Get yours from the dashboard.

​
Body

​
text
string
required
The text to ingest. Any length — automatically chunked by SemanticChunker.Best results: Use direct declarative sentences with named subjects.
✅ "Project Apollo uses PostgreSQL."
❌ "We use PostgreSQL." — pronoun subjects are filtered out.
​
user_id
string
required
Namespace identifier. Always overridden server-side from your API key — pass any non-empty string.
​
session_id
string
Optional sub-namespace for session isolation. Facts ingested with a session_id are only visible to searches using the same session_id.
​
persona
string
default:"shared"
Memory persona namespace within your org. Useful for separating memory by agent role (e.g. "engineering", "support").
​
source
string
default:"user"
Origin tag for provenance tracking.
One of: user agent doc slack email
​
use_llm_extraction
boolean
default:"true"
Use LLMGraphTransformer for knowledge graph extraction. Set to false for spaCy-only extraction — ~3× faster but lower recall.

​
Response

​
facts_ingested
integer
Number of knowledge graph triples stored in Neo4j.
​
episodic_chunks
integer
Number of vector chunks stored in Qdrant.
​
entities_extracted
integer
Unique entity nodes created or updated in the knowledge graph.
​
triples_extracted
integer
Raw triples extracted before deduplication.
​
latency_ms
float
Total processing time in milliseconds.
curl --request POST \
  --url https://api.atlas.bsyncs.com/brain/ingest \
  --header 'X-API-Key: atlas_your_key_here' \
  --header 'Content-Type: application/json' \
  --data '{
    "text": "Project Apollo uses PostgreSQL on AWS RDS.",
    "user_id": "user-123"
  }'

{
  "facts_ingested": 1,
  "episodic_chunks": 1,
  "entities_extracted": 2,
  "triples_extracted": 1,
  "latency_ms": 980
}