GitKraken Desktop
GitLens for IDEs 
Kepler
GitKraken MCP
GitKraken CLI
GitKraken Insights
Git Integration for Jira
Dev Team Automations
AI & Security Controls
Supercharge Your Dev Team 
Secure Your Dev Team 
X
Have questions or want to start a free trial of GitKraken Desktop On-Premise?
Your team can experience GitKraken Desktop On-Premise with all of its features, plus onboarding & support, 100% free of charge. We’re here to answer any questions about the product or pricing. In the meantime, we’ll create a trial license key and send you the details so you can take it for a test drive.
How GitKraken 3x'd Its Own Engineering Output by Building the AI Measurement Tools No One Else Had
Featuring perspectives from
Anastasia Zamyshlyaeva, VP of Engineering, GitKraken
Challenge
GitKraken needed to understand how deeply developers were adopting AI and turn that visibility into greater engineering output, but existing tools couldn’t provide the measurement or guidance its team needed.
Solutions
Key Outcome
Achieved nearly 3x engineering output without a larger budget. Gained actionable visibility into which AI workflows were delivering real productivity gains.
GitKraken's own engineering team hit 3.5x productivity gains, and unlocked an estimated 97 hours per developer, after building the AI-measurement and automation tools the market didn't have yet.
GitKraken’s engineering team is now shipping software at 3.5 times the productivity of a year ago, worth an estimated 97 hours of extra developer time per week, without asking for a bigger budget.
GitKraken got there by refusing to buy an off-the-shelf answer. When existing tools couldn’t tell VP of Engineering Anastasia “Stasia” Zamyshlyaeva whether her own developers were really adopting AI, or what to do about it if they weren’t, she and her team built two products instead: GitKraken Insights, which measures how developers actually use AI, and Kepler, which automates the work once they do.
Company
Industry
SaaS
Size
50-100 employees
Headquarters
Scottsdale, AZ
There was no solution designed for the AI-first world. The traditional metrics we had access to were vanity metrics, they weren't promoting the right type of behavior.
Anastasia Zamyshlyaeva, GitKraken VP of Engineering
Why Nothing Off the Shelf Worked
The push to become an “AI-first” engineering organization didn’t start as an abstract ambition. It came from real pressure: from the need to lead the market as a developer-tooling company rather than follow it, from the board of directors, and from a landscape changing week to week as new AI agents and models shipped.
When GitKraken committed to that shift, Stasia went looking for tools to measure it. What she found were software engineering intelligence (SEI) tools built for a pre-AI world. None of them could answer the questions that actually mattered: were developers adopting AI, and at what depth. The data that did exist arrived from disconnected sources, anonymized and stripped of context, with no roadmap attached for what a team should do next.
The available metrics were also the wrong ones. Lines of code and pull request counts measure activity, not capability. Put the wrong number in front of a developer, and the developer optimizes for the number instead of the outcome. GitKraken needed a way to measure AI adoption that reinforced the right behaviors instead of the easiest ones to game, gamification in service of real values rather than vanity numbers.
Kepler ADE (agentic development environment) integrates tasks and agents with your Git repository.
Where Kepler Came From
Rather than solving this from the top down, Stasia challenged one of her own engineers to go further than anyone else on AI adoption. That engineer was Giordano “Gyo” Piazza, our Director of Engineering, leading the team that builds Kepler.
The first week did not go well. By his own account, the sheer volume of what AI made possible was overwhelming. By week two, something shifted: Gyo could see how much faster the work could move, and became one of the team’s most energized adopters. That same week, he hit a problem none of GitKraken’s existing tools solved, managing the context an AI agent needs without giving it too little or burying it in too much. So he built his own prototype to solve it.
When he showed that prototype to the rest of the team, the reaction was immediate. Stasia brought it into a meeting of GitKraken’s product leaders, and the response made clear this wasn’t a one-person fix, it was the seed of a product. That prototype became Kepler, GitKraken’s agent-orchestration tool, now built out with a visual “agent graph” that lets developers watch what multiple AI agents are doing at a glance instead of reading a scrolling wall of log text, and an action system for stepping in at the exact moment a decision needs a human.
The Attempts to Get Measurement Right
Kepler solved the automation half of the problem. In 2022, the first generation of Insights gave developers direct access to raw engineering metrics inside GitKraken Desktop. The team quickly learned those numbers mattered more to engineering leadership than to the individual developers looking at them day to day, they were leadership-level signals dressed up as a developer tool, and adoption reflected that mismatch.
That lesson pushed the team to level up. The second generation, which launched in October 2025, moved to metrics built on DORA, an industry-standard framework engineering leaders already trusted for measuring software delivery. This version was driven by research from two GitKraken product leaders: Jeremy Castile, VP of Developer Research, and Jeff Schinella, Director of Product. Internal surveys had already validated what they suspected the market was asking for, and a Gartner report pointed in the same direction, while new entrants like DX, LinearB, and Span.app signaled a real shift underway in the space. Jeremy and Jeff paired that quantitative signal with direct interviews with engineering leaders, qualitative detail on the actual challenges those leaders were facing, and used both together to shape what GitKraken built next.
But that shift ran into a market that wouldn’t hold still. As AI tools multiplied, engineering leaders were being asked to adopt them immediately, then justify their impact almost as quickly, without the runway that real change management and training actually require. DORA and other traditional frameworks, built for a pre-AI world, weren’t built to answer those questions either.
That gap produced the third and current generation of Insights, purpose-built for AI-first engineering: agnostic of which AI provider or ticketing system a team uses, and built on the developer trust GitKraken had earned across the two generations before it. Justin, who leads GitKraken’s product team for Agentic Development (the group responsible for how AI agents fit into a developer’s actual day-to-day workflow), helped connect this new generation of Insights directly to how Kepler works in practice.
GitKraken Insights provides actionable insights, in addition to tracking agent impact, code quality, DORA metrics and more.
Two Sides of the Same Coin
Insights and Kepler started as separate answers to the same underlying problem: how do you lead engineering through an AI transformation nobody has a playbook for yet. GitKraken now treats them as two halves of one system, deliberately.
Kepler becomes so much more powerful compared to the rest of the market if you combine it with the data. And Insights becomes so much more powerful if you get the relevant data you're not able to get through any other means, and make it actionable through Kepler, where the developer is living.
Anastasia Zamyshlyaeva, GitKraken VP of Engineering
Insights provides visibility, into sessions, into where AI usage is paying off and where it isn’t, into which workflows are actually changing developer output versus which ones just look busy. Kepler acts on that visibility: feed it what Insights knows about which models or workflow steps produce results, and it can recommend the right move in the moment, not after a retro.
What Happens Next
Stasia is clear about where this goes next, and it maps directly onto GitKraken’s broader position on AI-driven development: automation happens under human guidance, it doesn’t happen instead of it.
With AI, part of that work can be automated, not replaced, under the guidance of a human. Otherwise you're guessing at things, or missing obvious gaps that should be improved.
Anastasia Zamyshlyaeva, GitKraken VP of Engineering
Every engineering team ships more than the product its customers see. It also continuously builds and refines its own development process, the internal workflow that determines how fast and how well that product actually gets made. Stasia’s argument is that this internal workflow deserves the same measurement rigor and continuous iteration as anything the company ships externally.
In that frame, Insights monitors the internal workflow the same way a product would be monitored: visibility into AI sessions, alerts when something drifts off track, a build-measure-learn loop applied to how the engineering org itself operates. Kepler is the automation layer inside that same loop, bringing the right context at the right time so a single developer can orchestrate five, ten, even fifteen parallel threads of work instead of two or three.
Behind the Numbers
The 3.5x productivity figure isn’t a one-off snapshot, it’s the result of a deliberate, company-wide push GitKraken ran on itself before ever selling the idea to customers. What the number reflects is the effort and complexity of what actually shipped to production, GitKraken’s own weighted view of delivery, not lines of code written and not pull requests opened.
In Q2 2025, the company ran an internal AI hackathon, a company-wide push to get employees hands-on with AI tools, no assigned use case, just a mandate to go try it. An “AI Olympics” followed, turning that early hackathon energy into an ongoing, team-based push rather than letting it fade after one event week.
By the end of February, that momentum became policy. GitKraken instituted a formal AI-first mandate: every team, every role, needed to be using AI. Not a pilot, not an opt-in experiment, a baseline expectation for how the company worked going forward.
On Stasia’s own team specifically, comparing a recent week to the month before the mandate (January), the productivity gain was close to 3x. By her estimate, that’s the equivalent of an extra 97 hours per developer, real budget the team didn’t have to go back to the board to ask for.
The data also functions as a roadmap. By studying how power users and regular adopters actually work, GitKraken can see what it takes to bring developers in the “emerging” and “explore” categories up to that level, and is using that pattern to decide what to invest in next.
Download Kepler ADE
Kepler is the agent-agnostic development environment for running AI agents across the entire development workflow, from idea or issue to merged PR. Get it for free!
Get A Demo Of GitKraken Insights
Book your demo for a one-on-one walkthrough of GitKraken Insights featuring real data and actionable engineering intelligence.
See What Your Own Team's Data Would Show
We didn’t build Insights and Kepler as a thought experiment. We built them because we needed real answers about our own AI adoption, whether developers were actually using it, where it was paying off, and where it wasn’t. Most engineering teams are asking the same questions right now with none of the same visibility.
GitKraken Insights shows you where that gap is: which teams and workflows are getting real value from AI, and which ones are just generating activity. Kepler puts that visibility to work, giving developers a single place to run multiple AI agents without losing control of the result.
Not replaced. Elevated. See what changes when your team can measure it.
Otherwise you're guessing at things, or missing obvious gaps that should be improved.
Anastasia Zamyshlyaeva, GitKraken VP of Engineering
About GitKraken
GitKraken is the AI-powered developer experience platform that fits any developer, any team, any budget. The platform spans developer tools like GitKraken Desktop, GitLens, and Automations with engineering intelligence through GitKraken Insights, giving both developers and engineering leaders what they need to build better software, faster.
Explore GitKraken Insights and try Kepler for free
Download Kepler ADE
Kepler is the agent-agnostic development environment for running AI agents across the entire development workflow, from idea or issue to merged PR. Get it for free!
Get A Demo Of GitKraken Insights
Book your demo for a one-on-one walkthrough of GitKraken Insights featuring real data and actionable engineering intelligence.
Add to Chrome
Add to Firefox
Add to Microsoft Edge