Purpose → Plan → Preparation
Use this file as your daily/weekly command center. Every initiative links back to its purpose, current plan, and preparation status.
- Start with Purpose — why does this matter?
- Check the Pipeline — which operational pipeline does this belong to?
- Check the Plan — what are we doing about it?
- Update Preparation — what's the current status and next action?
Pipelines are defined in docs/research/OPERATING_SYSTEM_FRAMEWORK_002-real-world-action-plan.md:
- Pipeline 1: Customer Discovery
- Pipeline 2: Content & Authority
- Pipeline 3: Product Build
- Pipeline 4: Pilot & Sales
- Pipeline 5: Operations & Infrastructure
- Pipeline 6: Founder Decision-Making
Status legend:
- 🟢 Ready / Active
- 🟡 In Progress / Blocked
- 🔴 Not Started
- ✅ Complete
Purpose: Unlock the highest-leverage disconnected asset — 1,501 lines of cited cognitive architecture.
Plan: Wire src/cognitive/ into the backend API.
Preparation:
- Read
src/cognitive/architecture - Map integration points to
backend/app.py - Design API endpoints for cognitive reasoning
- Write tests
- Deploy and validate
Status: 🔴 Not Started
Owner: @shivaramgoud
File: docs/tasks/prd/ARCHITECTURE_MAP_001-systems-and-bottlenecks.md → B5.1
Purpose: backend/app.py is 1,652 lines and violates single responsibility.
Plan: Decompose into modules: routes, services, middleware, config.
Preparation:
- Audit current endpoints
- Design module structure
- Refactor incrementally without breaking existing API
- Add tests
Status: 🔴 Not Started
Owner: @shivaramgoud
File: ARCHITECTURE_MAP_001-systems-and-bottlenecks.md → B2.1
Purpose: Contradiction detection is only 25% on synthetic ground truth.
Plan: Improve Step 5 prompts, test stronger models, build same-project corpus.
Preparation:
- Build parallel pipeline (19-20x speedup)
- Create synthetic contradiction corpus
- Collect same-project multi-discipline documents
- Rewrite Step 5 contradiction prompt
- Test GPT-4 / fine-tuned model
- Re-run TEST-002 report
Status: 🟡 In Progress
Owner: @shivaramgoud
Files: scripts/memo-poc/TEST_REPORT_001.md, real_construction_docs_contradictions/
Purpose: all-mpnet-base-v2 is not construction-tuned.
Plan: Evaluate/fine-tune embeddings on construction specs.
Preparation:
- Research construction embedding datasets
- Benchmark current retrieval quality
- Fine-tune or switch to domain model
Status: 🔴 Not Started
File: ARCHITECTURE_MAP_001-systems-and-bottlenecks.md → B3.2
Purpose: Current domain knowledge hardcodes US codes (IECC/IBC/ASHRAE); Dubai DM codes missing.
Plan: Integrate Dubai DM Building Regulations and GCC codes.
Preparation:
- Acquire Dubai DM regulations PDFs
- Build Dubai-specific code parser
- Validate against sample Dubai projects
- Add to domain knowledge base
Status: 🔴 Not Started
File: ARCHITECTURE_MAP_001-systems-and-bottlenecks.md → B4.3
Purpose: Build authority and reach in construction tech / VDC / AI circles.
Plan: Post 3x/week using the social media playbook.
Preparation:
- Write 10 post templates
- Create 14-day content calendar
- Set up Postiz scheduler
- Connect LinkedIn to Postiz
- Schedule first 3 posts
- Engage daily for 15 min
Status: 🟡 In Progress (blocked on LinkedIn API keys; see 90-day OS)
Files: docs/tasks/prd/SOCIAL_MEDIA_PLAYBOOK_001.md, postiz/, docs/research/OPERATING_SYSTEM_FRAMEWORK_002-real-world-action-plan.md
Purpose: Convert profile views into connections and conversations.
Plan: Update LinkedIn headline, banner, About section, featured section.
Preparation:
- Update headline
- Create banner image
- Rewrite About section
- Add featured items (Medha landing page, test report, architecture map)
Status: 🟡 In Progress (do in Week 1 of 90-day OS)
File: docs/tasks/prd/SOCIAL_MEDIA_PROFILE_OPTIMIZATION.md, docs/research/OPERATING_SYSTEM_FRAMEWORK_002-real-world-action-plan.md
Purpose: Position Medha as the context-preserving intelligence layer for fragmented construction platforms in GCC.
Plan: Research target companies, build outreach sequences, schedule calls.
Preparation:
- Build target list of 50 Dubai VDC agencies / contractors
- Write consultative outreach templates (Mom Test style)
- Identify warm intros via IIT Madras / Sub-engineering network
- Track outreach in CRM/spreadsheet
- Complete 15 customer discovery calls in 90 days
Status: 🟡 In Progress (Week 2+ of 90-day OS)
Files: MEDHA_CUSTOMER_MEETING_PLAYBOOK.md, docs/research/OPERATING_SYSTEM_FRAMEWORK_002-real-world-action-plan.md
Purpose: Synthetic contradictions prove the pipeline works; real contradictions prove market value.
Plan: Gather same-project documents with known conflicts.
Preparation:
- Source publicly available construction bid sets
- Request access to university project manuals
- Extract contradictions manually for ground truth
- Re-run pipeline and measure detection rate
Status: 🔴 Not Started
File: real_construction_docs_contradictions/
Purpose: Every architectural decision must be cited.
Plan: Keep docs/research/ and docs/decisions/ up to date.
Preparation:
- Cite new papers as they are used
- Write ADRs for major decisions
- Maintain BibTeX/Markdown bibliographies
Status: 🟡 In Progress
Files: docs/research/, docs/decisions/, docs/citations/
Purpose: One place to track all Medha work across Purpose → Plan → Preparation.
Plan: Self-host Plane and migrate tracker items into it.
Preparation:
- Deploy Plane on port 8091
- Open firewall rules (UFW + Azure NSG)
- Create Medha workspace
- Create projects: Product, GTM, Research, Operations
- Import tasks from
MEDHA_MASTER_TRACKER.md - Set up weekly review cycle
Status: 🟡 In Progress
Access: https://plane.trayini.ai
File: plane/README.md
Purpose: Prevent credential leaks and unauthorized access.
Plan: Rotate exposed keys, audit secrets, improve practices.
Preparation:
- Rotate OpenAI Key A (
sk-proj-VHFX...) - Rotate OpenAI Key B (
sk-proj--htNVT...) - Audit repo for any remaining exposed secrets
- Set pre-commit hook to scan for secrets (TruffleHog already configured)
Status: 🔴 Not Started (HIGH PRIORITY)
File: postiz/SECURITY_AUDIT_001.md
Purpose: Keep services running and accessible.
Plan: Monitor Postiz, backend, landing page, and agent bridge.
Preparation:
- Postiz deployed and accessible
- Azure NSG rules added
- Set up uptime monitoring
- Configure backups for Postiz data
Status: 🟡 In Progress
File: postiz/README.md
Purpose: Move from manual operations to agent-assisted pipelines that save founder hours while keeping human approval gates.
Plan: Deploy n8n, build MCP servers, ship first working agent (Synthesis Agent), then Content Agent, Ops Agent, Orchestrator.
Preparation:
- Answer discovery questions in
docs/research/agentic-enterprise-discovery/ - Deploy n8n self-hosted on Azure VM
- Evaluate existing Plane MCP servers
- Build
filesystem-mcpserver - Build
backend-mcpserver - Build Synthesis Agent (call notes → summary)
- Build Content Agent (draft → approve → Postiz)
- Build Ops Agent (backups, uptime, SSL, secrets scan)
- Build Orchestrator Agent (Sunday scorecard + drift alerts)
Status: 🟡 In Progress (discovery structured)
Files: docs/research/AGENTIC_ENTERPRISE_STACK_001.md, docs/research/agentic-enterprise-discovery/
Purpose: Experimental agent mesh has file-based shared memory and no frontend integration.
Plan: Decide whether to stabilize or deprioritize PicoCloth.
Preparation:
- Review PicoCloth current state
- Decide: integrate, rewrite, or archive
- If integrate: design frontend bridge
Status: 🔴 Not Started
File: ARCHITECTURE_MAP_001-systems-and-bottlenecks.md → B7.2/B7.3
These need a yes/no decision before proceeding:
| Decision | Options | Recommended | Status |
|---|---|---|---|
| Which enhancement track first? | A) Backend refactor, B) Frontend polish, C) MeMo hardening, D) Dubai corpus, E) PicoCloth stabilize | C) MeMo hardening | Pending |
| Should Postiz be on a subdomain? | Yes (postiz.drmath.trelolabs.com) / No (keep IP:4007) |
Yes | Pending |
| Should PicoCloth be archived? | Yes / No / Integrate | Archive for now | Pending |
| Target first customer segment? | VDC agencies / GCs / MEP subs / Owners | VDC agencies | Pending |
| Agentic stack first agent? | Synthesis / Content / Ops / Outreach | Synthesis Agent | Pending |
Answer these every morning:
- Purpose check: What is the single most important thing to move Medha forward today?
- Plan check: Which tracker item am I working on?
- Preparation check: What is the next concrete action I can take in the next 90 minutes?
- Blocker check: What am I waiting on or avoiding?
| Pipeline | Leading Signal | This Week | Blocker |
|---|---|---|---|
| Customer Discovery | Calls completed | __ / 2 | |
| Content & Authority | Posts published | __ / 3 | |
| Product Build | Real doc sets ingested | __ | |
| Pilot & Sales | Proposals sent | __ | |
| Operations | Backups + uptime | __% | |
| Founder Decisions | Scorecard complete | yes / no |
See docs/research/OPERATING_SYSTEM_FRAMEWORK_002-real-world-action-plan.md for stage-by-stage ownership and drift signals.
| Resource | Path |
|---|---|
| Architecture Map | docs/tasks/prd/ARCHITECTURE_MAP_001-systems-and-bottlenecks.md |
| Social Media Playbook | docs/tasks/prd/SOCIAL_MEDIA_PLAYBOOK_001.md |
| First 3 Posts | docs/tasks/prd/SOCIAL_MEDIA_FIRST_3_POSTS.md |
| LinkedIn Profile Optimization | docs/tasks/prd/SOCIAL_MEDIA_PROFILE_OPTIMIZATION.md |
| Postiz Setup | postiz/README.md |
| Security Audit | postiz/SECURITY_AUDIT_001.md |
| MeMo Test Report | scripts/memo-poc/TEST_REPORT_001.md |
| Customer Meeting Playbook | MEDHA_CUSTOMER_MEETING_PLAYBOOK.md |
| Research Landscape | docs/research/memo-analysis/CONSTRUCTION_AI_RESEARCH_LANDSCAPE_2026.md |
| Pipelined OS Action Plan | docs/research/OPERATING_SYSTEM_FRAMEWORK_002-real-world-action-plan.md |
| Agentic Enterprise Stack | docs/research/AGENTIC_ENTERPRISE_STACK_001.md |
| OS Bibliography | docs/citations/OPERATING_SYSTEM_FRAMEWORK_002.bib |
| Agentic Stack Bibliography | docs/citations/AGENTIC_ENTERPRISE_STACK_001.bib |
| Agentic Enterprise Discovery | docs/research/agentic-enterprise-discovery/README.md |
| Discovery Bibliography | docs/citations/AGENTIC_ENTERPRISE_DISCOVERY_001.bib |
Copy this into a new section each week:
### Week of [DATE]
**Wins:**
-
**Blockers:**
-
**Next week's focus:**
-
**Key metric:**
-