Platform
Five memory layers give agents context without turning every task into a retrieval problem. A memory gate decides whether retrieval is needed at all.
Current task state, plan, artifacts and active subtasks.
Durable facts and learned knowledge.
Past experiences, actions, outcomes, failures and corrections.
Reusable how-to knowledge, represented as skills.
Approval chains, handoffs, recurring processes and exception patterns.
Memory gate
Can current context handle the task?
│
Yes → Continue
│
No
↓
What knowledge is missing?
↓
Retrieve the correct memory /
knowledge / source of truthReduces noise, latency, token usage and stale-memory risk.
Model strategy
The model layer stays replaceable — local, on-prem or cloud. Specialization comes from personas, skills, tools and policies rather than from separate deployments.
Shared Base Model ├── Accountant ├── QA Engineer ├── Developer ├── Researcher └── Personal Assistant
Trace / experience ledger
Learning requires an auditable execution record. Runs are reproducible from versioned references.