# M1 — Reactive Copilot **Priority:** P0 **Target date:** ~April 1, 2026 (Wednesday) **Duration:** ~6 days from March 26 **Source:** [MOM 20260326](../phase-2/conversations/20260326-mom.md#act-iii--m1--m2--m3-milestones-locked-visualization-framework-priorities-0032--0057) --- ## Objective Deliver a **reactive copilot** with base visualization patterns and the redesigned UI connected to the Invictus/Tata Steel data source. This is the foundational user-triggered Q&A mode that orchestrates multiple agents to synthesize answers with visual outputs. --- ## What Ships ### Functional Deliverables - **Reactive copilot mode:** User asks question → system orchestrates agents → returns visual + textual answer - **Multi-agent orchestration:** Queries routed through RT core agent to data source agents (not single-source) - **Visual outputs:** Interactive charts, drill-downs, not just text responses - **Redesigned UI:** New patterns from design team integrated (code, not Figma) - **Invictus/Tata Steel data source:** Connected, queryable, trained data agent ### Technical Components - RT core agent functioning as hub: - User identity management - Question mode detection (ad hoc, training, data refresh) - Routing to user-aware and data source agents - Data source agent skeleton ready: - Manual training possible at developer level - Entity-specific agents (Invictus, Cautilia) can be created - Base visualization patterns operational --- ## Known Gaps (Accepted Technical Debt) 1. **Orange box missing:** - PostgreSQL database for system prompts/instructions not built - Current state: prompts hardcoded in Python - Impact: Less flexible, harder to update agent instructions - Mitigation: Deliver with embedded strings; refactor post-M1 2. **Orchestration wiring incomplete:** - Broader orchestration being stitched together - Architectural databases not fully set up - Impact: "Plumbing" delivery, not polished architecture - Mitigation: Focus on functional demo, not production-grade infrastructure 3. **Deferred RT core capabilities:** - Cross-source snapshot synthesis - RBAC for daily org data storage - Integration with two red boxes + one pink box - These are explicitly moved to M3 --- ## Parallel Workstreams ### Workstream 1: UI Integration **Owner:** TBD (UI team + Yeshwanth) **Tasks:** - Integrate redesigned UI code from Nihar's design team - Code must be ready (not Figma mockups) - Connect UI components to RT core agent API - Implement base visualization patterns (charts, tables, etc.) - Test interactive drill-downs and chart interactions **Dependencies:** - UI code delivery from design team - API contracts from backend team --- ### Workstream 2: Data Source Preparation **Owner:** TBD (Data team + Yeshwanth) **Tasks:** - Prepare Invictus/Tata Steel data set - Train data source agent on Invictus schema - Verify data agent can answer domain-specific queries - Test multi-agent routing (user query → RT core → data agent → response) - Validate visual output generation from data queries **Dependencies:** - Access to Invictus data - Data agent training framework operational --- ## Definition of Done ✅ **Functional:** - [ ] User can ask a business question through the UI - [ ] System routes question to appropriate data source agent(s) - [ ] Response includes both text explanation and visual component (chart/table) - [ ] Interactive elements work (drill-down, hover, filter) - [ ] Invictus data queries return accurate results ✅ **Technical:** - [ ] RT core agent successfully routes queries to data source agents - [ ] User-aware agent tracks identity and preferences - [ ] Data source agent responds to Invictus-specific queries - [ ] Redesigned UI code integrated and functional - [ ] Base visualization patterns render correctly ✅ **Documentation:** - [ ] Communication modes document updated (multi-agent orchestration + visual outputs) - [ ] API contracts documented for UI ↔ RT core agent - [ ] Known technical debt logged for post-M1 resolution --- ## Success Criteria **Demo scenario:** 1. CXO logs into Enterprise Brain 2. Asks: _"What are the top 3 production bottlenecks in Invictus this month?"_ 3. System: - Routes query to Invictus data agent - Agent queries data, identifies bottlenecks - RT core synthesizes response - UI renders: text summary + bar chart + drill-down table 4. CXO clicks on a bottleneck → detailed view appears **Quality bar:** - Response time < 10 seconds for simple queries - Visual accuracy matches data source - UI responsive and intuitive - No crashes or error states during demo --- ## Risks & Mitigations | Risk | Likelihood | Impact | Mitigation | | ---------------------------------- | ---------- | -------- | ---------------------------------------------------------------- | | UI code delivery delayed | Medium | High | Daily check-ins with design team; fallback to basic UI if needed | | Data agent training incomplete | Medium | High | Scope down to 3-5 key query types; expand post-M1 | | Orchestration wiring breaks | Low | Critical | Dedicated integration testing before demo | | Orange box absence blocks progress | Low | Medium | Already mitigated; hardcoded prompts acceptable for M1 | --- ## Post-M1 Cleanup Once M1 ships, immediately address: 1. Refactor hardcoded prompts into orange box (PostgreSQL) 2. Complete orchestration wiring for production 3. Set up architectural databases properly 4. Document technical debt for M2/M3 sprints --- ## Related Work **Todos from MOM:** - [x] Update communication modes document (multi-agent orchestration + visual outputs) — Yeshwanth - [ ] Grant Yeshwanth access to UI pattern code repository — This week - [ ] Ask Nihar's team to share UI generation prompts with Yeshwanth — This week **Related milestones:** - [M2 — Dynamic Workspace](M2-dynamic-workspace.md) — Builds on M1 with AI-driven UI - [M3 — Full Experience Center](M3-full-experience-center.md) — Completes vision with proactive mode **Related docs:** - [User Interaction Modes](user-interaction-modes.md) — Reactive vs proactive framing - [Data Flow Understanding](../phase-2/data-flow-understaning.md) — Architecture context --- **Last updated:** March 27, 2026