See how your developers are actually using AI — cost, security, compliance, productivity, and best practices.
One place to see AI spend, security exposure, compliance evidence, and delivery quality — across every team.
Overview
All teams · updated 09:40ExportAI spend · 30 days
$30.5k
▲ 26% vs prior 30 days
Driven by agent runs in Data Services
Open security reviews
7
2 unassigned
Oldest open record: 6 days
Evidence coverage
84%
▲ 9 pts since Q1
6 controls · Q2 pack ready
Rework rate
5.2%
▼ 0.4 pts vs prior 8 weeks
Changes reverted within 14 days
Month to date $30.5kAlert at $34k
Why now
Your engineers already build with AI. Leadership still can't see its cost, risk, or quality.
The console
One console gives leadership five connected views of AI in engineering: cost, security, compliance evidence, productivity, and a weekly team review.
What are we actually spending on AI tools, and why did it jump last month?
Compass AI shows spend by team, tool, and model, and flags the exact task behind last month's spike.
Can I prove nobody pasted a customer record or an API key into a chatbot?
Every flagged prompt lands in a triage queue with an owner, the surrounding context, and a resolution.
Audit wants proof our AI policy is enforced — not just a policy document.
Export the list of active rules and every issue they've caught, on demand.
My teams say AI made them faster. I can't tell if review load and rework got worse.
Adoption sits next to review load, rework rate, and test signals, every week.
I don't want to ban AI, and I don't want to spy on my engineers. What's the third option?
A weekly review assigns one action per team — expand, standardize, coach, or restrain.
| Team | Tools | Model | Tasks | Spend | Signal |
|---|---|---|---|---|---|
| Platform | Copilot, Claude Code | GPT-5.6 Terra | 1,204 | $11,800 | Steady |
| Payments | Cursor, ChatGPT | GPT-5.6 Sol (long-context) | 860 | $24,900 | Spike |
| Data Services | Claude Code | Claude Sonnet 5 | 640 | $9,500 | Steady |
Sample data.
Sample data.
| Rule | Category | Status | Issues caught | Last reviewed |
|---|---|---|---|---|
| Block secrets in prompts | Security | Active | 38 | Jul 14 |
| Redact customer PII | Data protection | Active | 21 | Jul 10 |
| Approved models only | Governance | Active | 9 | Jul 16 |
| Log external-model calls | Audit trail | Active | 52 | Jul 12 |
| Require review for sensitive repos | Access control | Active | 14 | Jul 08 |
| Retain prompt audit trail | Audit trail | Active | 8 | Jul 17 |
AI usage health · this month · 6 of 42 engineers shown
Sample data.
Weekly review · one action per team · no ranking
Team-level actions. No ranking, no surveillance of individuals. Sample data.
Time to value
From install to impact in about four weeks.
Onboard Compass AI
Connect your AI tools and teams.
Stop leaks on day one
Out-of-the-box rules catch secrets and policy breaks immediately.
Baseline cost & productivity
See real spend and delivery signals per team.
Cut cost, lift productivity
Act on the weekly review, in about four weeks.
~4 weeksDeploy Compass AI in days. Control cost and stop security and compliance leaks immediately.
Works for any engineering org — from a few developers to a few hundred.