Phase 1 complete — shipped

The Product Context Graph
for your IDE.

PIA (Product Intelligence Agent) builds a persistent intelligence layer inside your JetBrains IDE — connecting customer feedback, YouTrack tickets, and your codebase. Engineers know why. PMs prioritize with data. The graph compounds with every sprint.

See it in action ↓ How it works
ACP-compatible YouTrack + JetBrains Runs locally

The problem

Context-switching eats your sprints.

Engineering leads spend 30–40% of sprint planning re-building context that should already be attached to the ticket.

Archaeology every Monday

You open a ticket and spend 20 minutes in Slack and git history trying to remember why it exists and what was already tried last quarter.

No map of the codebase

The ticket says "fix the export bug." You have 40 service files and no idea which three actually matter for this change.

Scattered customer signals

Customer urgency lives in Intercom. Usage data is in Amplitude. Neither is visible when you're staring at a YouTrack ticket deciding what to prioritise first.

How it works

Three steps. One brief.

1

Type a ticket ID

In the JetBrains AI chat

Type any YouTrack ticket ID in natural language. PIA recognises it — no commands, no configuration, no forms.

PROJ-4521
2

PIA gathers context

Across three sources at once

Fetches the ticket and its linked issues, scans your local codebase for relevant files, and synthesises everything with Claude.

YouTrack API Codebase scan Claude AI
3

Get your enriched brief

Right inside the IDE

Business context, relevant files with inline hints, linked issues, customer signals, and suggested first steps — one response, zero tab-switching.

typically in under 3s

Interactive demo

See PIA in action.

Click a ticket — this is real mock data from the Phase 1 integration tests.

JetBrains IDE — PIA Agent Chat
Connected
Try a ticket:
◉

Select a ticket above to see PIA enrich it

Type a YouTrack ticket ID…

mock data · no real API calls · powered by Phase 1 test fixtures

Roadmap

Built layer by layer.

Each phase adds a new layer to the Product Context Graph. Phase 1 is live — the graph has started forming.

Phase 1
Phase 2
Phase 3
Phase 4
Phase 5
✓ Complete Q1 2026

Phase 1

Ticket Enricher

The foundation — connect, scan, synthesise.

  • ACP server (JetBrains IDE)
  • YouTrack connector
  • Local codebase scanner
  • LLM synthesis via Claude
  • Linked issues & comment history
Upcoming Q1 2026

Phase 2

Codebase Awareness

Replace keyword matching with LLM-powered understanding.

  • LLM-powered ticket-to-code linking
  • Git log analysis — co-change patterns & ownership
  • Module boundary detection
  • Complexity estimation from history & test coverage
  • Entry point identification for new work
Planned Q2 2026

Phase 3

Customer Demand Layer

Close the loop between customers and code.

  • Intercom signal connector — link threads to tickets
  • Amplitude usage data — who uses the affected feature?
  • Feature request clustering across accounts
  • Customer impact scoring — severity × account value
  • Revenue-weighted priority — make the business case for you
Planned Q2 2026

Phase 4

Product Context Graph

Full graph assembly — every entity connected, every claim traceable.

  • Persistent SQLite graph: customers → feedback → tickets → code → outcomes
  • Evidence chains — every insight traceable to raw data
  • Cross-source entity linking (YouTrack + Intercom + Sentry + Amplitude)
  • Graph survives across sessions — context compounds over time
Planned Q2 2026

Phase 5

Intelligence Layer

"What should we build next?" — answered with data.

  • Sprint prioritization scored by ARR impact & implementation risk
  • Post-ship impact tracking — did the feature actually work?
  • Data-backed sprint planning — not whoever argued loudest
  • Defensible prioritization rationale for every sprint decision