Persistent graph memory for ChatGPT Desktop. MCP-native. Two-minute install.
ChatGPT's built-in memory stores short, opaque summaries. You cannot query it, traverse it, or trust its provenance. Graphory gives ChatGPT a structured graph: typed nodes, typed edges, and a full audit trail of every fact.
Plugs into ChatGPT Desktop's MCP server slot. No plugins, no wrappers, no custom GPTs.
ChatGPT can traverse real business relationships instead of fuzzy-matching summary blobs.
No embeddings, no hallucinated recall. Queries return exact results from your graph.
Every memory carries a timestamp, a source, and an audit trail. ChatGPT can answer when and where.
Works with any OpenAI model you use in ChatGPT. Graphory is the memory layer, not the reasoning layer.
Install the SDK, log in, point ChatGPT Desktop at the MCP server.
Step 1 - Install and authenticate
pip install graphory graphory login
Step 2 - Register Graphory in ChatGPT Desktop's MCP config
{
"mcpServers": {
"graphory": {
"url": "https://api.graphory.io/mcp",
"headers": {
"Authorization": "Bearer gs_ak_your_key"
}
}
}
}Illustrative snippet. ChatGPT Desktop's MCP config path and exact schema vary by release - check the ChatGPT Desktop settings panel or OpenAI docs for the current location.
Real queries ChatGPT can run once Graphory is connected.
"Remember that Sarah from Acme mentioned pricing was their blocker in March."
"Show me everything I learned about Project X across all my ChatGPT sessions this quarter."
"Who have I interacted with at companies in the fintech vertical?"
Deterministic extraction and retrieval, benchmarked against public datasets.
0.9107 accuracy
LongMemEval (agent + MCP)
0.8667 accuracy
LoCoMo-MC10 (agent + MCP)
0.900 F1
BizLineItemBench