2,700
companies analyzed
$2.7T
in enterprise value
19
AI adoption metrics in one model
13
metrics from source control alone
AI Adoption
See what AI is changing in your engineering
Liz connects to source control and shows where AI is in use, how AI work compares to non-AI work, and what that means for the business. Folks are using it for AI adoption and impact, keeping roadmaps on track, and automating board reports.
Built on experience
Sema has analyzed tech companies for investors for 9 years
The logic model
From AI spend to business outcomes
19 metrics, read left to right. Resources are what the organization brings. Activities are what AI changes. Outputs are what the work produces. Outcomes are what the organization gets, and they matter most.
Resources · What you bring
3 metrics: the pre-AI baseline, AI tool seat adoption, and seat activity versus observed commits.
Activities · What AI changes
Writing code with AI tools, and setting up the repository for AI. These are the intervention, not metrics.
Outputs · What the work produces
12 metrics. AI in use: share of commits and pull requests, breadth, tooling and readiness. AI versus non-AI: review time, size, reverts and cycle time.
Outcomes · What the organization gets
4 metrics: defect escape rate, features shipped, lead time from ticket to ship, and cost per merged pull request. Each is measured against the pre-AI baseline.
Example · anonymized
Edtech company: 85% of code AI adjusted
From 30 minutes of setup.
85%
of committed code is AI-assisted.
+36%
pull requests.
~40%
more features shipped, counted in the work tracker.
0
change in the regression rate.
Two depths
Start with 1 connection. Add 3 more when ready.
Depth is about data access, not performance. A Depth 1 organization can score well on every metric it collects.
Depth 1 · AI in use
Source control only. Always on. 13 of the 19 metrics: where AI is used, and how AI work compares to non-AI work.
Depth 2 · Payoff
6 more metrics, including all 4 outcomes. Work tracking adds 3, AI tool telemetry 2, finance 1.
Beyond engineering
AI adoption in go-to-market
See how broadly and frequently go-to-market teams use AI, and the variety of workflows it supports. The report gives sales, marketing and customer-success leaders a shared view of adoption.
How it works
Three steps
Step 1 · Connect source control
Liz reads your history to set the pre-AI baseline.
Step 2 · See AI in use
13 metrics fill in: AI in use, and AI versus non-AI.
Step 3 · Add connections
Work tracking, AI tool telemetry and finance unlock the other 6, including all 4 outcomes.
Questions
Frequently asked
Does Liz rate individual engineers?
No. Liz reports at the team level, never the individual.
What do you need to start?
Read access to source control. Everything else is optional.
Can the results go to the board?
Yes. Folks use Liz to automate board reports on AI adoption and impact.
Get started
See your AI adoption, one connection in
Connect source control and we will walk you through the result live.
