Roadmap

The public roadmap focuses on the surfaces external teams use to integrate Neutron AI.

Current Focus

• Backwards-compatible REST APIs for memory, temporal entities and relationships, graph queries, context, decisions, consequences, outcomes, learning review, simulations, agents, contradictions, snapshots, and hybrid search.
• TypeScript SDK facades for the complete intelligence surface and authenticated REST guidance for SDKs awaiting typed parity.
• A focused MCP surface for memory, graph, context, decision, outcome, and simulation workflows.
• Product-proof journeys for bitemporal history, multi-graph reasoning, Nucleus isolation, Context Engine permissions, Pareto trade-offs, approvals, outcome learning, provider boundaries, and Durable Object eviction.
• Public documentation for Nucleus and Scope design, provenance, confidence, snapshots, model-independent solvers and simulators, deletion, and reviewed learning.

Next

• Typed knowledge and decision parity for Python, PHP, Go, and Rust SDKs.
• Admin and platform workflows for graph exploration, contradiction review, decision snapshots, outcome comparison, and learning activation.
• Public benchmark reports with versioned latency, cache, and quality metrics.
• Calibrated domain simulation and optimisation integrations with explicit validation and proof boundaries.
• Migration guides for each breaking API or SDK change.

Later

• More granular token scopes.
• Signed webhook events for integration observability.
• Richer memory, graph, decision, and audit export workflows.
• More evaluated examples for support, coding, agriculture, finance, workplace, research, and operations agents.