Skip to content

vittobh/pm-os

Repository files navigation

pm-os — AI-Native PM Operating System

Author: Vittobha Vignesh S · Senior PM · AI / BFSI / Healthcare

About

pm-os is a Product Operating System that converts a Product Requirements Document into a full software development lifecycle. It orchestrates Claude Skills, MetaGPT, and Promptfoo so that a Product Manager, working alone, can move from a PRD to an eval-ready SDLC package in under two hours. Built by a Senior Product Manager for teams shipping AI features in regulated industries.

Live application: https://vittobh.github.io/pm-os/


Live: vittobh.github.io/pm-os Status: Active · v0.4


What is pm-os?

pm-os is a full-lifecycle PM operating system that takes a product from first idea to post-launch review — generating every artifact a PM needs along the way.

It does not replace Jira, Confluence, SharePoint, or Google Workspace. It orchestrates them, adding a PM-specific workflow layer on top of Atlassian Rovo, Microsoft 365 Copilot, Google Workspace AI, and Claude.


The Problem

Product managers spend 60–80% of their working week on documentation and re-work:

  • Market research, SWOT, competitor analysis, ROI models — produced from scratch every time
  • Jira, Confluence, SharePoint, Yammer, Google Drive searched separately with no unified view
  • AI exists in each tool (Rovo, M365 Copilot, Gemini) but is not connected into a PM workflow
  • Nothing is reused across products, teams, or client engagements

Two-Tier Product Model

Standard AI ✨
Search Federated KB search: Jira + Confluence (Rovo) · SharePoint + Yammer + Teams (M365) · Google Drive + Docs (Google API) Same + AI chat grounded in KB (Rovo Chat + Claude)
Reporting Embedded PowerBI + Looker standard templates PowerBI Copilot + Looker AI — natural language queries and screenshot interpretation
Research Browse + reuse artifact URLs Deep research: internal KB + web, structured report (Quick / Standard / Deep)
Artifacts PM templates library — clone and edit AI-generated drafts: SWOT, PRD, ROI model, KPI plan, release strategy
Org Multi-PM org KB · Client workspace (read-only) AI-summarised client status · cross-product pattern recognition

Four Engines

1. Discovery Engine

Input: problem statement + domain Output: market sizing (TAM/SAM/SOM) · SWOT · competitor matrix · ROI model · compliance risk map · GTM brief

2. Delivery Engine

Input: PRD Output: prioritised backlog (RICE/WSJF/MoSCoW) · KPI plan · Y-Score gate · beta strategy · phased rollout plan

3. KB Layer

  • Indexed across: Jira, Confluence (Rovo) · SharePoint, Yammer, Teams (M365) · Google Drive, Docs (Google API)
  • Every artifact tagged: domain · stage · product-type · compliance-regime
  • Standard: search + reuse URLs · AI: Rovo Chat + Claude, cited responses
  • Client workspace: PM creates, client views — no Atlassian/M365 licence required for client
  • Shared Library: anonymised templates reusable across all client engagements

4. Code Pipeline (original pm-os)

Input: PRD → MetaGPT (architecture) → OpenHands (code) → Promptfoo (QA) → Cloudflare (deploy)


Live Pages

Page URL
Landing vittobh.github.io/pm-os
PM Onboarding vittobh.github.io/pm-os/onboarding
CuCP Slide Deck vittobh.github.io/pm-os/cucp/presentation.html
Teams Tab vittobh.github.io/pm-os/onboarding/teams.html

Repository Structure

pm-os/
├── prds/                          # Reference PRDs
│   ├── 00-pm-os-platform.md       # Master platform PRD (v0.4)
│   ├── 01-y-score-framework.md    # Launch readiness framework
│   ├── 02–08-*.md                 # Domain reference PRDs
│   └── 09-cucp-program-ops.md    # Coupa CuCP 90-day rollout
├── docs/                          # GitHub Pages (live site)
│   ├── index.html                 # Landing page
│   ├── onboarding/
│   │   ├── index.html             # 6-step PM onboarding wizard
│   │   ├── teams.html             # Microsoft Teams entry point
│   │   └── manifest.json          # Teams app manifest
│   ├── cucp/
│   │   └── presentation.html      # CuCP slide deck (6 slides)
│   └── product/
│       ├── SRS.md                 # System requirements (v0.4)
│       └── FRD.md                 # Functional requirements (v0.4)
├── ROADMAP.md                     # Phased delivery plan
└── .github/workflows/pages.yml   # Auto-deploy to GitHub Pages

Integration Strategy

pm-os builds on existing enterprise AI — it does not reinvent it:

Layer Tool Used pm-os Adds
Jira + Confluence search + AI Atlassian Rovo (Search + Chat) PM templates, KB tagging, cross-product patterns
SharePoint + Yammer + Teams Microsoft 365 Copilot / Graph API Unified view alongside Atlassian
Google Drive + Docs + Chat Google Workspace API + Gemini Same unified view, Google-native orgs
PowerBI + Looker reports PowerBI Embed + Copilot · Looker API PM pre-built templates, AI narrative layer
AI chat + artifact generation Rovo Chat + Claude Sonnet 4.6 PM workflow context, cited KB responses

Constraints

  • LLM cost < $0.50 per full discovery pack (Claude Sonnet 4.6 with prompt caching)
  • All outputs are human-editable markdown — no lock-in
  • AI outputs cite every source with URL (EU AI Act Art. 13 explainability)
  • Standard tier: zero LLM cost — Rovo + M365 + Google search only
  • Client data is fully isolated — separate tenants, never cross-contaminated
  • Rovo / M365 Copilot / Google Workspace licences required at org level

Getting Started

git clone https://github.com/vittobh/pm-os.git
cd pm-os
# Set API credentials
export ANTHROPIC_API_KEY=your_key
export CLOUDFLARE_API_TOKEN=your_token
# Run orchestrator against any PRD
bash orchestrate.sh prds/00-pm-os-platform.md

Or open the live onboarding: vittobh.github.io/pm-os/onboarding


Built With

Atlassian Rovo · Microsoft 365 Copilot · Google Workspace AI · Claude API (Anthropic) · MetaGPT · OpenHands · Promptfoo · Cloudflare Workers


Built by Vittobha Vignesh S · Senior PM · AI / BFSI / Healthcare

About

A PM's operating system: turn a PRD into a full SDLC by orchestrating MetaGPT, OpenHands, Promptfoo. Built by a Senior PM for regulated AI (BFSI · Healthcare).

Topics

Resources

License

Code of conduct

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Packages

 
 
 

Contributors

Languages