# Uniph.ai Roadmap

Last updated: February 2026

This roadmap reflects what is already shipped, what is currently in progress, and what is next by execution priority.

---

## Guiding Principle

Uniph.ai is an objective-first AI life orchestration platform: autonomous by default, governed at risk boundaries, and measured by outcome quality.

---

## Shipped

- Core LifeOS surfaces: users, goals, tasks, integrations, infer
- Collaboration core: workspaces, contributions, runs, events, summary/questions/context/outcome
- Agent management: registration, API keys, workspace permissions, identity records, identity audit
- Identity health telemetry: reconnect/staleness detection via `GET /api/agents/:id/identity/health`
- Governance and execution primitives: contracts, policy evaluation, approvals, action jobs, trace
- Memory v1: create/list/recall/stats/compact/retention/purge
- Friction-first onboarding: `POST /api/orchestrate/start` plus KPI endpoints
- Channel adapters: Slack, Discord, Email
- Continuous self-improvement baseline: feedback signal ingestion, regression summaries, release-gate checks

---

## In Progress

### 1) Agent identity OAuth completion

- Provider-native token exchange and refresh lifecycle
- Hardened revoke/disconnect semantics
- Better provider health and reconnect UX (health endpoint + reconnect prompts now live)

### 2) Memory quality and lifecycle

- Higher precision recall and stale-memory suppression
- More robust scheduled compaction and retention automation
- Safer destructive workflows and user-facing controls

### 3) API and integration DX hardening

- Clarify core orchestration, governance, memory, and adapter API entry paths
- Improve identity lifecycle docs and reconnect/recovery flows
- Reduce integration setup time-to-first-outcome

### 4) Mission-control and evaluation UX

- Surface high-level outcomes, risk state, and quality gates without trace overload
- Tighten onboarding/evaluation harness usability for operator checks
- Preserve deep trace access for audits and debugging when needed
- Add Improve panel feedback loop (regressions, active experiments, release posture)

### 5) Release reliability gates

- Environment and migration verification in deployment
- Automated smoke gates and rollback guardrails
- Expanded operational dashboards

---

## Next (ordered)

1. Complete provider-native identity OAuth for Google/Microsoft first.
2. Improve memory quality controls (edit/forget/compact/retention) with stronger evaluation baselines.
3. Finalize API DX pass for orchestration/governance/memory/integrations.
4. Ship clearer mission-control summaries aligned to approvals-only user intervention.
5. Enforce release gates in CI/CD to prevent production regressions.
6. Expand security and data governance (retention, PII policy, export/delete completeness).

---

## Later Expansion Tracks

- Deeper model routing and cost/reliability optimization
- Richer governance policy authoring and approval workflows
- Stronger mission-control analytics and scenario harnesses
- More integration adapters and ecosystem recipes

---

## Success Signals

- Lower time-to-first-outcome and fewer required user actions
- Higher autonomous resolution rate without trust regressions
- Improved memory precision and lower stale recall rate
- Higher API onboarding completion and faster adapter activation
- Fewer deployment regressions and faster recovery when failures occur