Skip to content

Examples

Start local. Move to runtime when another process needs to consume the same graph contract.

1. Local Graph Memory

from seocho import Seocho, Ontology, NodeDef, RelDef, Property

ontology = Ontology(
    name="accounts",
    nodes={
        "Company": NodeDef(properties={"name": Property(str, unique=True)}),
        "Product": NodeDef(properties={"name": Property(str, unique=True)}),
    },
    relationships={
        "PROVIDES": RelDef(source="Company", target="Product"),
    },
)

client = Seocho.local(ontology, llm="mara/MiniMax-M2.5")
client.add("Beta provides risk analytics to ACME.")

print(client.ask("What does Beta provide?"))

2. Index Multiple Records

client.add_batch([
    "ACME acquired Beta in 2024.",
    "Beta provides risk analytics to ACME.",
    "ACME uses the risk analytics product for compliance review.",
])

print(client.ask("How is Beta related to ACME?", reasoning_mode=True))

3. Use A Running Runtime

from seocho import Seocho

client = Seocho(base_url="http://localhost:8001", workspace_id="default")

result = client.raw_ingest(
    [
        {"id": "doc-1", "content": "ACME acquired Beta in 2024."},
        {"id": "doc-2", "content": "Beta provides risk analytics to ACME."},
    ],
    target_database="accounts_graph",
)

print(result.status)

4. Query With Bounded Repair

semantic = client.semantic(
    "What is ACME related to?",
    databases=["accounts_graph"],
    reasoning_mode=True,
    repair_budget=2,
)

print(semantic.response)

5. Direct Runtime API

curl -sS -X POST http://localhost:8001/platform/ingest/raw \
  -H "Content-Type: application/json" \
  -d '{
    "workspace_id": "default",
    "target_database": "accounts_graph",
    "records": [
      {"id": "doc-1", "content": "ACME acquired Beta in 2024."}
    ]
  }'

Next