Runtime and conversations
Agent execution, streaming, bounded history, delegation, retries, token caps, tracing, and cooperative cancellation.
Read runtime architectureOpen development roadmap
Roadmap
See what is available now, what is being built next, and which safety and governance foundations come before higher-risk capabilities.
Early-stage public build. Useful foundations exist, but production suitability must be assessed rather than assumed.
Available now
Foundations you can use today
The current build provides the runtime, governance, tool, file, scheduling, and audit foundations for governed agents.
Governance and operations
Workspace identity, roles, invitations, approval controls, schedules, audit and security views, and side-effect envelopes.
Read governance architectureTools and skills
A typed tool registry, one audited dispatch path, tool catalog, web search, skill documents, runtime loading, and management UI.
Review completed plansFiles and storage
Provider-neutral storage, file revisions, processing jobs, file tools, management UI, scratch space, and multimodal input.
Review completed plansNext and later
Capabilities being added next
Current work expands integrations, organisational knowledge, persistent memory, operational monitoring, and safer code execution.
Integration foundations
Credentials, OAuth, resource discovery, active context, first provider packages, and their management surfaces.
Open the roadmapEvaluation and tool hardening
Behaviour scenarios, bounded tool results, calibrated token estimates, and stronger protection around untrusted content.
Open the roadmapKnowledge and persistent memory
Governed ingestion, hybrid retrieval, evaluation, agent tools, editable memory, and visible product surfaces.
Open the roadmapArtifacts and public launch
Versioned outputs, sandboxed serving, share controls, community health, supply-chain work, and the first formal public release.
Open the roadmapArchitectural gates
Higher-risk capabilities have prerequisites
Integrations, memory, retrieval, and external actions only move forward when the controls around them are ready.
Side effects need eyes
Operational governance must exist before integrations ship tools that can spend money or change external systems.
Assemble context once
Skills, files, integrations, knowledge, and memory must compose through one budgeted prompt-assembly design.
Evaluate before tuning
Retrieval, memory, and harness behaviour need scenarios and expected outcomes before optimisation changes land.
Treat external content as untrusted
New content sources require a threat model and adversarial fixtures before their output reaches model context.
Follow the work
Explore detailed milestones on GitHub
Follow completed work, upcoming capabilities, and the technical decisions shaping each phase.