SEOCHO | Ontology-Aligned Graph Memory

Ontology-aligned memory OS

Memory that keeps its shape.

SEOCHO turns ontology, graph facts, semantic queries, and agent runtime into one contract. Teams can inspect what an agent knows, why it knows it, and which relationships it is allowed to use.

Understand SEOCHO Run the quickstart GitHub

Ontology allowed shapes

Graph stored evidence

Query bounded retrieval

Runtime agent access

Brand promise

A graph memory system with visible rules.

SEOCHO should feel closer to infrastructure than a demo app: compact, inspectable, and built around contracts that survive across local experiments, production runtimes, and open-source contribution.

Why this exists

Contract

Ontology first

Define the entity and relationship vocabulary before agents start writing memory.

Memory

Graph native

Persist facts, provenance, and constraints as relationships that can be inspected.

Answer

Evidence bound

Route questions through ontology and graph context before response synthesis.

Runtime

Agent ready

Expose the same contract through SDK calls, HTTP endpoints, traces, and policy checks.

Product paths

Start from the work in front of you.

The site should not ask readers to decode the repository first. Each path names the job, the next page, and the part of SEOCHO it exercises.

Understand

Read the map

Start with the concept guide: ontology, graph memory, query path, runtime, and ownership.

Concept guide 02

Build

Create local memory

Use Seocho.local(...) to define one ontology, add one fact, and ask one grounded question.

Quickstart 03

Operate

Run the contract

Move to the runtime when multiple agents need the same HTTP surface, traces, and policy checks.

Runtime docs 04

Contribute

Shape the project

Use the public issue, release, community, and PR rules to keep changes small and reviewable.

Playbook

First useful run

Prove one memory loop locally.

The shortest path should feel calm: define the ontology, add one record, ask one question, and inspect the graph contract before touching deployment.

Quickstart Python SDK Examples

copy-paste local path Python

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

ontology = Ontology(
    name="work",
    nodes={
        "Person": NodeDef(properties={"name": Property(str, unique=True)}),
        "Company": NodeDef(properties={"name": Property(str, unique=True)}),
    },
    relationships={
        "WORKS_AT": RelDef(source="Person", target="Company"),
    },
)

client = Seocho.local(ontology, llm="mara/MiniMax-M2.5")
client.add("Marie Curie worked at the University of Paris.")
print(client.ask("Where did Marie Curie work?"))

System shape

Four surfaces, one contract.

Ontology contract

valid node, relationship, and property shapes

Graph memory

facts, provenance, constraints, and repair artifacts

Semantic query

ontology-aware retrieval before final answers

Runtime boundary

HTTP, traces, policy checks, and shared workspaces

Docs map

Field notes

Concepts, reasoning, and major release notes.

All posts

Latest 2026-05-01

SEOCHO Design Philosophy & Operating Principles

Extract domain rules and high-value semantics from heterogeneous data into a SHACL-like semantic layer.

2026-04-19

Turning a Folder of Filings into a Compliance Graph in Under a Minute

A runnable SEOCHO usecase — declare a 6-entity compliance ontology, ingest six mock filings, and ask questions that cross regulator / incident / control / policy boundaries without writing Cypher.

2026-02-21

Feasibility Review Framework & Rubrics

Multi-role feasibility review framework and Go/No-Go rubric for graph data implementations.

Project signal

Latest updates

Release-linked updates from the public GitHub trail.

Commits

2026-06-13 Sync integration established. Awaiting first release. #init