We build 50x faster with AI.Your dev team can too.
Simply connect GitHub or GitLab to Proxon Productivity and see how AI is lifting your engineers, where work is getting stuck, and the next steps to go faster. Plus six months of history, ready in minutes.
Meet the 100x engineer.
A 10x engineer is about talent. A 100x engineer is about systems: one person directing a team of agents, with the tools and safeguards to trust what they produce. The best engineers are publishing their numbers, and those numbers keep climbing.
Peter Steinberger, who built OpenClaw, runs 5 to 10 agents at once across 48 repositories, about 118 commits a day. He spent much of his time building the tools that make those agents better.
Lauren Tan says volume was never the goal: verify first, then run agents in parallel. August ended at 2,462 pull requests in production.
OpenAI's team driving Codex wrote no code by hand: about 3.5 merged pull requests per engineer per day, in roughly a tenth of the usual time.
We didn't just predict the shift. We lived it.
Proxon runs on Proxon. Same team as a year ago, building in a completely different way: agents write the code, engineers direct the work, and the output compounds week after week. The biggest leap so far is in the build phase. Planning, review and operations are next.
- engineering output vs. a year ago, with the same team
- 50x
- pull requests merged in 30 days by our lean team
- 690
- median time between merges
- 16 min
- of the work we merge is AI-assisted
- 96%
- AI-assisted
- Human only
You can't lead a shift you can't see.
You're paying for AI coding tools, and leadership wants to know what they're getting. GitHub and GitLab can't tell you how much of your work AI is doing, or whether you're on the path the leading teams are on. We've done the work to know which questions to ask, and how to answer them.
- GitHub tells you how many pull requests merged, not how many AI helped write.
- Getting an answer means custom API queries, a chatbot, and knowing which questions to ask.
- AI usage and spend live in one place, engineering work in another.
- AI-assisted and other work side by side: output, cycle time and review, team by team.
- No single score and no verdict. You decide what good looks like for your team.
- Sits next to Adoption and Cost Intelligence, so the whole AI picture is in one place.
See where AI is lifting your team, and where to push next.
Connect GitHub or GitLab and Productivity matches your pull request history to the AI usage Proxon already tracks. It's built around three questions every engineering leader is being asked.
How much more are we getting done with AI?
How much of your merged work AI helped write, and how output and cycle time are moving, with AI and without.
Is AI creating new bottlenecks?
Whether AI work is skipping human review, how big its pull requests run, and where work is waiting.
What do we do next?
Flags on what's worth a look, like AI work skipping review or a team that hasn't started with AI, so you know where to act first.
We'll help your team make the climb.
We've made this shift ourselves. A Proxon expert comes on site, learns how your engineers use AI today, and trains them on a modern, AI-first way of building software. Then they hand it off, and Productivity keeps measuring your progress.
Talk to us about working with an expert →Baseline
We interview your team, connect GitHub or GitLab, and show where your team stands today.
Train
We teach the fundamentals, then workflows built for your team, with the review habits that keep AI work safe.
Hand off
Your internal lead owns the playbook. Productivity keeps tracking progress, so leadership sees the results.
Your baseline, in minutes.
Productivity connects to GitHub or GitLab directly. All you need is an admin to approve the connection.
Connect GitHub or GitLab
An admin approves the Proxon app for your organization.
Pick your repositories
Choose the repositories you want to track. You can add more any time.
Add Jira or Linear (optional)
Adds metrics like time from ticket to first commit.
See six months of history
Data shows up in seconds to minutes, so you start with a baseline instead of waiting weeks.
Already a Proxon customer? Productivity is in your sidebar. Sign in to see your numbers
Questions engineering leaders ask
Last updated
Usually within seconds to minutes of connecting GitHub or GitLab, depending on how many repositories you choose and how large they are. Proxon also loads up to six months of history, so you start with a baseline instead of waiting weeks for new data to build up. Nothing gets installed on anyone's laptop.
We gave AI agents fully specified work so each could run for 30 minutes or more alone, built tooling so many agents can work in parallel, and added safeguards like independent code review and hard gates. That's code-writing speed: planning, deploying and operating still move closer to human pace.
Productivity measures the AI share of merged pull requests, merges per engineer, and cycle time from first commit to merge, split into stages to show where the time goes. It also tracks pull request size and how much AI-assisted work merges with a human review, for AI work and everything else.
A pull request counts as AI-assisted when a commit has an AI co-author or an AI tool's attribution appears in the description. Pull requests an agent opened are shown separately, so you can compare them with the rest. Treat the share as a floor: inline completions and pasted code often leave no trace in the pull request.
Compare AI work with everything else. Look at how much AI work merges without a person approving it, how large its pull requests run, and which stage of cycle time is growing. If AI work tracks close to the rest, there is no bottleneck yet. A widening gap is the early warning.
No. It shows patterns across teams and over time, not proof of cause. Teams that use AI heavily may also differ in the work they take on, their seniority or their tooling. Use the numbers to decide where to look closer and what to try next, not as a final verdict.
Access is set by role. Admins decide who sees team views and who sees individual usage, with least-privilege defaults. Proxon reports on how teams use AI so leaders can spread what works. It is not a keystroke or content monitor.
Today Productivity covers engineering, because pull requests, reviews and merges are work every engineering team already measures. We're building the same AI and non-AI comparison for other teams, using tools like Linear. If you want productivity insights for a team outside engineering, book a demo and we'll show you what's possible.
Proxon works in three steps: visibility into how your company uses AI, insights from outside your company on what good looks like, and solutions that act on it. Productivity is mostly visibility today. It sits next to Adoption and Cost Intelligence, so engineering results and AI spend live in one place.
Yes. Productivity shows where AI is working and where it isn't. If you want help moving forward, a forward-deployed expert from our team assesses where you are, builds a roadmap for what's next, and trains your engineers on a modern, AI-first way of building software. Then Proxon measures your progress over time.
Yes. Productivity is the same product that was called Velocity, renamed so the name matches the question it answers. It measures the same things and uses the same connections. Only the name changed, to match what leaders actually want to know.
Build a team of 100x engineers. We'll show you how.
Book a demo and we'll connect your repositories and walk through what Productivity shows: where AI is already paying off, where it's getting stuck, and the next step for your team.
Book a demoAlready a Proxon customer? Productivity is in your sidebar. Sign in to see your numbers