title: "Autonomous Knowledge Agent — Daily Research & Intelligence System"
type: ticket
subtype: architecture_design
purpose: "Design and implement an automated knowledge-building agent that researches defined topics daily, builds a persistent knowledge base, and delivers structured intelligence briefings."
Design and deploy an automated research agent that:
| # | Topic | Scope | Priority |
|---|---|---|---|
| 1 | AI Technologies, Tools & Players | New models, frameworks, benchmarks, funding, acquisitions. Key players: Anthropic, OpenAI, Google, Meta, Mistral, xAI. Focus on agentic AI, MCP, tool-use advances. | P0 |
| 2 | Claude Ecosystem | Claude tools, skills, plugins, MCP servers, Claude Code updates, API changes, community projects, prompt engineering advances, Desktop/mobile features. | P0 |
| 3 | Odoo & Odoo.sh | New modules, integrations, v19 changes, community addons, Odoo.sh platform updates, OCA contributions, deployment patterns. | P1 |
| 4 | BIM (Building Information Modeling) | Standards evolution, software updates (Revit, ArchiCAD, Tekla), IFC developments, openBIM movement, digital twins, BIM mandates by country. | P1 |
| 5 | ISO 19650 | Standard revisions (2026 DIS for Parts 1&2), national annexes, certification bodies, compliance tools, industry adoption news. | P0 |
| 6 | Construction Industry Intelligence | Players & roles, pain points, tools & solutions, AI initiatives in AEC, market trends, regulatory changes, mega-projects, labor dynamics. | P2 |
Brain 3 must evaluate these storage strategies and recommend ONE primary + ONE backup:
STRUXIO_Logic/knowledge/
├── ai_technologies/
│ ├── STATE_topic_index.yaml # Topic maturity, last updated, source count
│ ├── foundation/ # Initial deep research articles
│ │ ├── 2026-03-18_anthropic_landscape.md
│ │ └── 2026-03-18_agentic_frameworks.md
│ └── daily/ # Daily deltas
│ ├── 2026-03-19_news.md
│ └── 2026-03-20_news.md
├── claude_ecosystem/
│ └── ...
└── BRIEFING_daily.md # Synthesized daily briefing across all topics
Pro: Git-tracked, grep-searchable, Claude can read directly, zero infra cost.
Con: No semantic search, manual retrieval, scales poorly past ~500 files.
STRUXIO_Logic/knowledge/ (Git-tracked, Claude-readable)1. TRIGGER: Manual "run research" command (later: cron at 06:00 IST)
2. FOR EACH TOPIC:
a. Load STATE_topic_index.yaml → check maturity level
b. IF maturity == "foundation":
- Deep web search (10-15 queries per topic)
- Build comprehensive overview articles
- Create foundation/ documents
c. IF maturity == "active":
- Targeted news search (3-5 queries, last 24h filter)
- Compare against existing knowledge (detect true deltas)
- Create daily/ news document (only genuine new information)
d. Update STATE_topic_index.yaml
3. SYNTHESIZE: Generate BRIEFING_daily.md across all topics
4. NOTIFY: Place briefing in Founder's morning queue
| Phase | Action | Assigned To |
|---|---|---|
| 1 | Read this ticket + RULE_brain_protocol.md | Brain 3 (Chat/Opus) |
| 2 | Evaluate architecture options, recommend primary storage | Brain 3 (Chat/Opus) |
| 3 | Define STATE_topic_index.yaml schema | Brain 3 (Chat/Opus) |
| 4 | Create directory structure + index files | Brain 3 → Code arm |
| 5 | Execute foundation research: AI Technologies | Brain 3 (Chat/Opus — web search) |
| 6 | Execute foundation research: ISO 19650 | Brain 3 (Chat/Opus — web search) |
| 7 | Write daily briefing template | Brain 3 → Cowork arm |
| 8 | Write runbook for manual trigger | Brain 3 → Cowork arm |
| 9 | Document DEVXIO migration path | Brain 3 (Chat/Opus) |
| 10 | Submit to Brain 1 for review | Brain 3 |
All output is additive (new files). No existing files modified. Rollback = delete STRUXIO_Logic/knowledge/ directory.