Dev Intelligence, resolved to real people.
Dev Intelligence ties version control, issue tracking, and AI usage into a single weekly read for engineering leaders. Cycle time, review load, and AI assist hours, resolved to real people, with verdicts instead of stoplights.
GITHUB · BITBUCKET · JIRA · CLICKUP · CURSOR · ANTHROPIC · OPENAI
THE TRIAGE MOMENT
The worst number finds you first.
Before any tile is read, the thing that needs attention is already legible.
-
Opinionated verdicts
Great, Good, Needs attention. The product takes a position so the read is not yours to invent.
-
Labeled semantics
Business hours, calendar hours, logged hours: the unit is part of the number, every time. Drift erodes trust faster than missing features.
-
Honest about gaps
When data is stale or missing, the dashboard says so. No empty-state cheerleading.
“We’ve given out GitHub Copilot to everyone… from an audit perspective, we’ve lost complete track.”
WHAT IT TRACKS
Built for the metric you'll defend in a 1:1.
Every number labels its semantics. Every chart names its source and window.
-
Cycle time p50
How long work actually takes to ship. The unit is part of the number: business hours, calendar hours, logged hours.
-
Review load p95
Who carries the review queue, and when concentration becomes a bottleneck. Two reviewers at 71% is a 1:1, not a curiosity.
-
AI assist hours
Where AI usage actually lands, across Cursor, Anthropic, and OpenAI. Adoption you can see, not survey.
-
AI spend & intensity
Total spend, spend per commit, spend per merged PR, lines accepted, and an intensity score per person. The cost side of adoption, not just the usage side.
-
Identity resolution
The same engineer across GitHub, Jira, and AI providers, resolved into one activity story.
-
Weekly pulse
One read per week with a verdict on top. Not forty charts you have to interpret yourself.
THREE OUTPUTS
Three ways to read the same source of truth.
The dashboard for managers, the API for the rest of the toolchain, MCP for the assistant that already lives in your editor.
-
DASHBOARD · FOR HUMANS
Concentrated in two reviewers. Worth a 1:1.
The weekly read: pulse tiles, verdicts, and fourteen days of activity. Built for the manager with ten minutes, not an analytics team.
-
REST API · FOR MACHINES
The same pulse, piped into your warehouse, board pack, or alerting.
-
MCP SERVER · FOR ASSISTANTS
Drop it into your Claude config and DevInt becomes a tool your assistant can call.
UNDER THE HOOD
Three sources in. Three outputs out. One engine.
HOW IT CONNECTS
Connected in a day. Reading in a week.
Connect
Read-only integrations: GitHub or Bitbucket, Jira or ClickUp, and your AI providers. No code changes, nothing installs on developer machines.
Resolve
Identity resolution ties accounts across providers to real people. One engineer, one activity story, no matter how many logins they have.
Read
A weekly pulse with a verdict on top, daily activity underneath. When something needs attention, it says so first.
WHO IT'S FOR
Three people ask three different questions. One answer.
-
ENGINEERING MANAGER
"Is my team healthy this week?"
The pulse is built for the metric you'll defend in a 1:1. Verdicts, labeled units, and the receipts behind every number.
-
CTO
"Which teams need help?"
The same read across every squad, vendor, and pod. Trendlines for the board without commissioning a data project.
-
CFO
"What does the AI spend do?"
You can see an engineer spent $400 on tokens yesterday. DevInt shows what the $400 produced: assist hours next to delivery output, in the same frame.
“You guys are at a more formalized, productized end of where we are headed.”
FAQ
What tech leaders ask about Dev Intelligence.
Version control: GitHub and Bitbucket. Issue tracking: Jira and ClickUp. AI usage: Cursor, Anthropic, and OpenAI. Three sources, one read.