Piper Morgan

An AI-powered product management assistant — and an experiment in human-AI collaboration.

Current version: v0.9.0.0 private beta · pipermorgan.ai · GitHub

What Piper Morgan does

Piper Morgan is a conversational PM assistant that connects to the tools a product manager already uses — GitHub, Notion, Slack, Google Calendar — and helps make sense of them together. Ask it what’s blocking the sprint, surface what needs attention today, or get a morning standup that says what’s actually happening rather than what the last person to update a ticket said was happening.

It’s built by a team of AI agents working alongside a human PM. That’s both the product and the point.

Private beta

Piper Morgan is currently in invite-only private beta at beta.pipermorgan.ai.

If you’re part of the beta:

• Beta Quick Start — get running in a few minutes
• Testing Guide — what to test and how to give feedback
• Known Issues — current limitations
• Terms — terms and expectations
• Release Notes v0.9.0.0 — what changed in the latest release

Not in the beta yet? You can follow along at pipermorgan.ai — we publish weekly ships, building narratives, and insights about what we’re learning.

Current capabilities

• Conversational interface — natural language for PM tasks; Piper tracks context across a conversation so you don’t repeat yourself
• GitHub integration — issue triage, status, prioritization, cross-feature patterns
• Morning standup — honest assembly from live sources, with explicit provenance (no hallucinated progress)
• Integrations — GitHub, Notion, Slack, Google Calendar; connector architecture (RECONNECT) actively extending
• Radar — structured display layer for the objects Piper tracks (WorkItems, Documents, People, Conversations)
• Trust-gating — proactive skills surface when invited; consequential actions require explicit confirmation
• Per-user key routing and field encryption — credentials isolated per user, secrets encrypted at rest (AES-256-GCM)

Architecture

FastAPI server (port 8001)
  ├── Intent dispatch → workflow rail
  ├── Integration services (GitHub · Notion · Slack · Calendar)
  ├── Radar object layer (WorkItem · Document · Conversation · People)
  └── Trust + skill routing (ADR-072)

PostgreSQL 14+ (port 5433) — primary store
Redis 7+ (port 6379) — session cache
ChromaDB (port 8000) — vector search
LLM: user-keyed — Anthropic Claude and/or OpenAI (every call bills the acting user's own key)

Key docs (working corpus — on GitHub, not this site):

• Architecture Decision Records — every architectural decision with rationale
• Patterns catalog — reusable implementation patterns
• BRIEFING-CURRENT-STATE.md — live sprint status (agents: read this first)

Running Piper locally

# 1. Clone and set up
git clone https://github.com/mediajunkie/piper-morgan-product.git
cd piper-morgan-product
python -m venv venv && source venv/bin/activate
pip install -r requirements.txt

# 2. Configure
cp .env.example .env
# Add your API keys (Anthropic, GitHub, etc.)

# 3. Start services and launch
docker compose up -d
alembic upgrade head
python main.py   # server at localhost:8001

Note: If launching from a Claude Code shell, strip the inherited Anthropic env vars first — see the warning in CLAUDE.md.

Where things live

The working corpus lives in the repository — this site serves the visitor-facing documentation; everything below is one click away on GitHub:

What’s next

Piper Morgan is in invite-only private beta (v0.9.0.0, cut 2026-10-31). The MVP milestone closed 2026-10-30. The release sequence from here: Production (required for the public release, cut as 1.0.0.0 when the Production milestone closes), then Fast Follow.

Follow the weekly ships at pipermorgan.ai/shipping-news to see what’s shipping.

Support

• Docs: pmorgan.tech (you’re here)
• Issues: GitHub Issues
• Beta support: support@pipermorgan.ai
• Blog: pipermorgan.ai