In production

The graph behind Wexa.ai.

One case study, told properly — because we'd rather show you a real workload than a wall of logos. CognoDB is the context graph inside Wexa.ai, the AI workspace it was originally built for.

The workload

Wexa.ai builds a persistent graph of a company's work — people, documents, tools, and the dependencies between them. Every document reference and workflow dependency is a node or a relationship, and agents traverse that graph in real time while answering questions and taking actions.

Why it's demanding

The graph is under continuous write pressure: four separate ingestion paths converge on one Bolt connection pool, feeding a live instance while agents read from it. Batched UNWIND … MERGE keeps re-ingestion idempotent, and query latency has to stay low enough to sit inside an agent's response, not behind it.

What it shaped

This workload is why CognoDB is built the way it is: fast instance startup because agent workloads spin graphs up per-session, sustained ingest throughput because memory updates never stop, and MCP as a first-class interface because the consumers are agents, not dashboards. The ingest and compatibility numbers on this site come from the benchmark suites this workload made necessary.

“Every relationship, document reference, and workflow dependency is stored as a graph node and traversed in real time as agents answer questions and take actions.”
Wexa.ai — AI workspace
  • Workload
    Personal & enterprise knowledge graphs
  • Write pressure
    Four ingestion paths, one Bolt pool
  • Access pattern
    Real-time traversal during agent runs
  • Interface
    Cypher over Bolt + MCP
The shape of an agent-side readcypher
MATCH (p:Person {email: $who})-[:WORKS_ON]->(proj)
MATCH (proj)<-[:REFERENCES]-(d:Document)
RETURN proj.name, collect(d.title)[0..5] AS docs

Your logo here, honestly

We'll add customers as they're real.

If you're running CognoDB in production and willing to be named, we'd love to tell the story with the same specificity — numbers, shapes, caveats and all.

Start now

~98.7%

token efficiency at 2,000 entities — see the footnotes above

Put your first graph up in a minute.

A free instance takes about a minute and no card. Write two MERGE statements, read them back, and you have a living graph — with provenance on every fact.

Start free View the quickstart

First-graph path

Live

  1. Create a free instance

    No card. Ready in about a minute.

  2. Connect your driver

    bolt+ssc:// URI into the driver you already use.

  3. Write two MERGEs

    That's the entire shape of agent memory.

  4. Point an agent at it

    One MCP config block. No integration code.