ACE
A context and retrieval engine for ranking, routing, and surfacing relevant information across complex AI workspaces and agent workflows.
Treats context as infrastructure: ranks, prunes, and routes sources so agents see only what is strictly necessary for the current step.
Problem Space
Most RAG systems fail because they provide too much irrelevant information, leading to token waste and model confusion.
System Direction
ACE ranks and prunes retrieval sources based on task-aware execution logs, ensuring the model only sees what is strictly necessary for the current step.
Public Capabilities
- 01Context resolution
- 02Source ranking
- 03Task-aware retrieval
- 04Lower token waste
- 05Reduced hallucination risk
ACE is preparing for open source release. The repository will be made public when the initial release is ready.
What Is Not Disclosed
Private implementation details, security-sensitive internals, and unreleased runtime architecture are intentionally not disclosed.