ProductGraph - Product Requirements Document¶

Moved from root PRD.md - See PLAN.md for current implementation status.

Executive Summary¶

ProductGraph is an AI-native runtime product intelligence platform that unifies design visualization, user journey analytics, and application observability into a single system. It replaces the fragmented toolchain of Figma (design), Amplitude/Pendo (analytics), and Datadog (monitoring) with a unified runtime graph that serves as both documentation and measurement system for how products actually perform.

Vision Statement¶

A living, visual map of user journeys that doubles as both documentation and a real-time measurement system for how well those journeys actually perform.

Problem Statement¶

Current Tool Fragmentation¶

Teams today use disconnected tools that each capture a partial view of the product:

Tool Category	Examples	Limitation
Design	Figma, Sketch	Static pixels, not runtime-aware
Analytics	Amplitude, Mixpanel	Abstract tables/charts, detached from UI
User Guidance	Pendo, WalkMe	Overlay system, manual event definition
Monitoring	Datadog, New Relic	Backend-focused, limited UI context
Session Replay	FullStory, LogRocket	Post-hoc debugging, no journey context

Core Problems¶

Design-Reality Gap: Figma shows intended flows; users experience something different
Manual Event Instrumentation: Analytics require developers to manually add tracking
No Unified Journey View: Funnels, debugging, and design exist in separate systems
AI Agents Lack Observability: Coding assistants can edit code but can't observe real UI behavior

Goals¶

Primary Goals¶

Unified Runtime Canvas: Visual graph of user journeys generated from actual application state
Automatic Event Capture: Component-level tracking without manual instrumentation
Spec vs Reality: Compare intended journeys with actual user behavior
AI Development Loop: Enable AI agents to observe, debug, and improve products

Non-Goals (v1)¶

Full Figma design replacement (focus on runtime, not static design)
Mobile app support (web-first)
Backend distributed tracing (focus on frontend journeys)

User Personas¶

1. Product Manager¶

Needs: Understand user behavior, identify drop-offs, validate features
Current Pain: Switching between Amplitude dashboards and Figma specs
ProductGraph Value: See journey success rates overlaid on visual flows

2. UX Designer¶

Needs: Validate designs with real user data, identify friction
Current Pain: Design specs don't reflect actual user paths
ProductGraph Value: Visual diff between intended and actual journeys

3. Frontend Developer¶

Needs: Debug UI issues, understand state transitions, fix user-reported problems
Current Pain: Reproducing user issues, correlating logs with UI state
ProductGraph Value: Step-by-step replay with component state at each step

4. AI Coding Agent¶

Needs: Observe UI changes after code edits, understand runtime behavior
Current Pain: Can edit code but can't see results; limited feedback loop
ProductGraph Value: Visual feedback on code changes, automated testing

5. Growth Engineer¶

Needs: Optimize funnels, A/B test UI changes, improve conversion
Current Pain: Manual funnel setup, delayed insights
ProductGraph Value: Automatic funnel detection, real-time metrics

Product Concepts¶

Runtime Canvas¶

A visual graph where each node represents a step in the user journey:

[Landing Page]
     │
     ├──60%──► [Product List]
     │              │
     │              ├──30%──► [Product Detail]
     │              │              │
     │              │              └──50%──► [Add to Cart]
     │              │                             │
     │              │                             └──40%──► [Checkout]
     │              │
     │              └──70%──► [Exit]
     │
     └──40%──► [Exit]

Each node contains: - UI snapshot - Component tree - State values - Performance metrics - Success rate

Journey Graph¶

The underlying data model connecting: - Pages: Route-based application screens - Steps: User actions (clicks, inputs, navigations) - States: Application state at each step - Metrics: Conversion rates, timing, errors

Requirements¶

See TRD.md for detailed technical requirements.

Success Metrics¶

Metric	Target	Timeline
Instrumented applications	5+	Q3 2025
Events ingested/day	1M+	Q3 2025
Journey insights generated	100+	Q4 2025
User drop-off identification accuracy	>90%	Q4 2025
Time to first insight	< 5 minutes	Q3 2025

Competitive Analysis¶

Capability	Amplitude	Pendo	Datadog	FullStory	ProductGraph
Event analytics	★★★★★	★★★☆☆	★★☆☆☆	★★★☆☆	★★★★★
Visual journeys	★★☆☆☆	★★★☆☆	★☆☆☆☆	★★★★☆	★★★★★
Session replay	☆☆☆☆☆	★★☆☆☆	☆☆☆☆☆	★★★★★	★★★★☆
Component state	☆☆☆☆☆	☆☆☆☆☆	☆☆☆☆☆	★★☆☆☆	★★★★★
AI integration	★★☆☆☆	★☆☆☆☆	★★☆☆☆	★☆☆☆☆	★★★★★
Auto-instrumentation	☆☆☆☆☆	★★★☆☆	☆☆☆☆☆	★★★★☆	★★★★★
Spec vs reality	☆☆☆☆☆	☆☆☆☆☆	☆☆☆☆☆	☆☆☆☆☆	★★★★★