Pensero MCP

Engineering intelligence, now queryable.

Query your live Pensero data straight from Claude, ChatGPT or any MCP-compatible assistant. Ask about delivery, quality, AI adoption or benchmarks in plain language.

Book a demo

Works with the assistants your team already uses

Claude
OpenAI
Gemini
GitHub Copilot

Ask it the way you’d ask your best analyst

Plain questions in, answers grounded in your own data out, then keep pulling the thread with follow-ups.

How did engineering perform this quarter?
Where is AI creating measurable impact?
Which teams are improving fastest?
What should I highlight in my board update?
How do we compare against benchmark companies?
Which engineering investments are delivering the highest ROI?

Why Pensero MCP

Four things that separate an answer from Pensero MCP from an answer a general-purpose assistant guesses at.

01 — Source of truth

Built on your engineering source of truth

Every answer is generated from your live Pensero data, the same unified model behind your dashboards. No exports, no stale spreadsheets, no numbers that disagree with the ones your team already reviewed.

Live data
One unified model
No exports

Total DeliveryTotal points delivered29.9kpts12%

Weekly Delivery Per PersonAverage weekly delivery per active contributor58.3pts9%

AI-written codeShare of merged code that was AI-written64%18%

Cost per pointAI spend per delivery point among active contributors$1.128%

Defect cost %Share of delivery spent fixing bugs9%6%

Roadmap AlignmentShare of delivery aligned with roadmap68%7%

Collaboration RatioShare of delivery spent collaborating21%4%

Issue Cycle Time (P90)P90 time from ticket assignment to PR merge96.4hrs12%

Total DeliveryTotal points delivered29.9kpts12%

Weekly Delivery Per PersonAverage weekly delivery per active contributor58.3pts9%

AI-written codeShare of merged code that was AI-written64%18%

Cost per pointAI spend per delivery point among active contributors$1.128%

Defect cost %Share of delivery spent fixing bugs9%6%

Roadmap AlignmentShare of delivery aligned with roadmap68%7%

Collaboration RatioShare of delivery spent collaborating21%4%

Issue Cycle Time (P90)P90 time from ticket assignment to PR merge96.4hrs12%

Pensero MCPLive connection to Claudeconnected

02 — Conversation

Natural conversations, not report requests

Start broad, then drill in. Ask why a number moved, which team it came from and what changed since last month, without leaving the assistant you already have open.

Follow-up questions
Anomaly investigation
Trend exploration

What should I highlight in my board update?

Comparing delivery since January

Lead with delivery. Delivery points per person are up 28% since January. The whole curve moved, so it’s the whole team, not a few people.

Jan – Feb · mean 244Jun – Jul · mean 312

0200400600

Delivery points per person, monthly

Mean 244 → 312Change +28%Spread flat

Which teams drove it?

Ranking teams by change

Biggest movers since January

Platform 18 people+41%

Payments 11 people+33%

Mobile 9 people+24%

Pensero MCPconnected

Claude

03 — Compatibility

Works with the tools you already run

Pensero exposes a standard MCP server, so it connects to Claude, ChatGPT and any MCP-enabled assistant. Nothing new for your team to learn, no extra tab to keep open.

Claude
ChatGPT
Any MCP client

04 — Security

Enterprise-ready by default

MCP uses your existing Pensero authentication and permissions. Every request is scoped to the person asking, so nobody can reach data they couldn’t already see in Pensero.

Existing authentication
Permission-scoped
No new access surface

"It's important that we select the right information at the right moment, based on the maturity and the knowledge that we have on the platform."

Jordi AbadCTO, SEAT CODE

Read story →

"I can go to Pensero and tie all this together in a great format."

Nick TraylorHead of Engineering, Stord

Read story →

Questions teams ask about MCP

Which assistants does it work with?

Any MCP-compatible assistant, including Claude and ChatGPT. If your assistant speaks MCP, it can connect to Pensero.

Who can see what?

Every request runs under the identity of the person asking, using your existing Pensero authentication and permissions. Nobody can reach data through MCP that they could not already see in Pensero.

Is the data live?

Yes. Answers are generated from your live Pensero data at the moment you ask — not from a snapshot, an export or a cached report.