Companies that use Github Copilot

We dug into our own data to find which companies are using GitHub Copilot in production. We also asked software engineers working in these companies to tell us how they're using it. Here are interesting examples of how the biggest companies in the world are using GitHub Copilot.

ABN AMRO

Banking · Amersfoort, Netherlands · GitHub Copilot

ABN AMRO is one of the biggest banks in the Netherlands. They do everything a large bank does: mortgages, savings accounts, business lending, and everything in between. Tens of thousands of employees serving millions of customers across Europe.

The interesting part is what's happening inside their mortgage business. There's a small machine learning engineering group of about four engineers who handle the less glamorous but critical work of taking AI models out of the lab and running them in production. They're the people who make sure a fraud detection model or a risk scoring model actually works reliably once real customers start hitting it.

This little group has been quietly rewiring how they work around GitHub Copilot. Rather than just using it as a fancy autocomplete, they're actively researching and implementing AI agents inside Copilot to automate pieces of their own development process. Things like picking up tickets, generating boilerplate code, and handling the routine plumbing that used to eat into their day.

The tooling around it is interesting too. They work in Python, lean on Azure DevOps for project tracking, use VS Code as the editor, and pipe large language models into various parts of the workflow. Essentially, the bank is using Copilot agents to speed up the work of the people who speed up the bank's own AI.

For a regulated institution, this is a notable posture. Rather than treating AI tools as something to cautiously pilot somewhere far from production, they've put them at the heart of the team whose whole job is shipping production machine learning for mortgages.

---

XP Inc.

Financial Services · São Paulo, Brazil · GitHub Copilot

XP Inc. is one of the biggest independent financial companies in Brazil. If you live there, you've probably heard of them through one of their brands like XP, Rico, or Clear. They help millions of people invest their money, from first-timers buying their first stock to serious investors managing large portfolios.

XP has a dedicated group focused on developer experience and AI, whose job is essentially to make the rest of engineering faster. They're not just handing out Copilot licenses and hoping for the best. They're building custom agents and automations on top of Copilot, integrating it into their pipelines, and tracking real metrics like how long it takes a change to go from idea to production.

What makes their setup genuinely interesting is how deep they go on the plumbing. They work with newer standards like Model Context Protocol and Agent Client Protocol, which are basically shared languages that let AI agents talk to tools and to each other in a consistent way. They also experiment with knowledge graphs, which give the AI a structured map of how XP's own systems and code fit together, so suggestions are grounded in reality rather than pulled from thin air.

The practical payoff is that a Copilot agent at XP isn't just autocompleting a line of code. It's something closer to a junior engineer that actually understands the shape of XP's codebase, can pick up tickets, and can hand work off to other agents or ask a human when it gets stuck.

---

PNC

Financial Services · Pittsburgh, Pennsylvania · GitHub Copilot

PNC is one of the largest banks in the United States. They're headquartered in Pittsburgh and serve millions of customers across the country with everything you'd expect from a big bank: checking accounts, mortgages, business lending, credit cards, and wealth management.

Banks this size run an enormous amount of software behind the scenes, and a huge chunk of it involves moving data around. Loan applications, transaction records, risk reports, fraud signals, customer analytics, all of it has to flow from one system to another reliably and safely. PNC has a whole part of the company dedicated to this called Data and Automation, and they've brought GitHub Copilot in as a core part of how their engineers work.

The stack is heavy on the data side. Engineers there work in Python, PySpark, and Hadoop to build pipelines that process the bank's data at massive scale. Copilot sits inside their editor helping them write and review that code, which for pipeline work is particularly useful because so much of it is repetitive plumbing that AI is genuinely good at accelerating.

What's interesting is that PNC has folded Copilot into their formal engineering practices rather than treating it as a side tool. Their engineers use it alongside established disciplines like code reviews, continuous integration and deployment, and automated testing. The goal is a faster delivery cycle without loosening any of the checks that matter when you're writing software for a regulated bank.

Essentially, PNC is treating AI-assisted coding as just another part of modern engineering, the same way version control or automated testing became standard a generation ago. That's a pragmatic move for an institution where the consequences of a bad release can show up in millions of customer accounts.

---

Optum

Healthcare · Bangalore, India · GitHub Copilot

Optum is the technology and services arm of UnitedHealth Group, one of the largest healthcare companies in the world. They handle the less visible side of American healthcare: pharmacy benefits, data analytics, care delivery, and the software that makes insurance claims and medical records actually work behind the scenes.

A company that size runs a lot of software, and a lot of it lives in the cloud. Optum has been moving large parts of its technology onto public cloud providers, and GitHub Copilot is a core part of how their engineers work during that shift. Rather than treating AI tools as a nice-to-have, Optum explicitly expects engineers to use them to move faster through the cloud migration.

The practical workflow is what makes it interesting. Engineers use Copilot to speed up the repetitive work of writing infrastructure code, the scripts that define how cloud services get deployed and configured. They use it to build and refine the automated pipelines that push new code into production. They also use it alongside Microsoft 365 Copilot for things like summarizing meetings and drafting documentation, so the AI assistance carries through from writing code to writing everything around the code.

What's notable is how deliberate Optum has been about the guardrails. The engineers working with Copilot are specifically walled off from protected health information and personally identifiable data. They're building and improving software, not touching real patient records. For a healthcare company where a data leak can mean serious regulatory consequences, keeping the AI-assisted engineering work cleanly separated from sensitive data is a meaningful architectural choice.

Beyond basic Copilot use, Optum engineers are also building their own AI products on top of services like OpenAI's APIs and Google Gemini, and experimenting with frameworks like LangChain and Microsoft Copilot Studio to create custom assistants. The AI isn't just helping them ship code faster. It's becoming part of what they ship.

---

Wolters Kluwer

Software · Alphen aan den Rijn, Netherlands · GitHub Copilot

Wolters Kluwer is a Dutch company that builds specialized software for professionals who deal with complicated rules for a living: tax accountants, lawyers, doctors, compliance officers, and financial auditors. If you've ever had a tax return prepared or a medical chart reviewed, there's a decent chance some Wolters Kluwer software was involved behind the scenes.

Software like that has to be extremely reliable. A tax tool that miscalculates or a medical reference that shows the wrong dosage isn't just annoying, it's genuinely dangerous. That means Wolters Kluwer puts a lot of effort into testing their products before customers ever see them, and that testing work is where GitHub Copilot has found a home.

Their test automation group uses Copilot specifically to help write the automated scripts that check their software behaves correctly. They work with tools like Playwright, which simulates a real user clicking through a web application, and Copilot helps engineers generate those test scripts faster. Given that a single product might need thousands of these tests covering every feature and edge case, the time savings add up quickly.

What's interesting is that Wolters Kluwer is also pushing into a newer area called agentic AI testing, where AI doesn't just help write the tests but can autonomously explore an application and figure out what to test on its own. That's genuinely cutting-edge work, and most companies are still theorizing about it rather than actually doing it.

There's also a cultural note worth mentioning. Wolters Kluwer has a firm rule that AI tools like Copilot can be used to build their software, but not during job interviews for people applying to build that software. They want to see what candidates can actually do themselves. It's a small detail, but it reflects a thoughtful stance: AI is a tool for the work, not a substitute for the person doing it.

---

Yelp

Technology · San Francisco, California · GitHub Copilot

Yelp is the reviews site almost everyone has used at some point, whether to find a decent restaurant, a plumber, or a hair salon. Behind the familiar review pages there's a substantial sales and support operation that actually makes money for the company, and that's where some of the most interesting engineering happens.

Yelp has a team dedicated to what they call telephony, which is the phone-based infrastructure that their sales and customer success representatives use every day to talk to business owners. This is where a lot of Yelp's revenue actually flows through, so the tools have to work reliably at scale.

What makes this genuinely interesting is the AI layer being built on top. Yelp is developing voice bots powered by large language models that can handle real customer calls, along with real-time call transcription and AI-generated summaries for sales reps. The goal is a shared voice AI platform that multiple Yelp products can plug into.

GitHub Copilot is part of the core toolkit engineers use to build all of this, sitting alongside a notably broad set of AI coding assistants including Claude Code, Cursor, and others. Yelp treats AI tooling as table stakes for the engineering environment rather than something exotic.

The practical result is that engineers there are writing React plugins for their call center software, building serverless functions on AWS, and integrating voice AI features, all with Copilot as a steady companion in the editor. For a company most people still think of as just a reviews site, there's surprisingly sophisticated voice AI infrastructure being built in the background.

---

Descartes Systems Group

Software · Waterloo, Ontario, Canada · GitHub Copilot

If you've ever ordered something online and wondered how it actually gets from a warehouse to your door, a lot of the invisible infrastructure behind that journey runs on software from companies like Descartes. They're one of the largest logistics and supply chain technology companies in the world, with more than 26,000 customers using their cloud systems to move goods across borders, through ports, and onto trucks.

Their research and development hub is working on something genuinely interesting: modernizing a large-scale logistics system that covers the entire lifecycle of a shipment. That means everything from the moment a package enters the system, through customer assignment, routing, and final delivery, plus invoice generation and reporting at the end.

GitHub Copilot is at the center of how this team builds. They work in Visual Studio on a Microsoft stack that includes .NET, ASP.NET, and SQL Server, and they've integrated Copilot as an active AI assistant in daily development. The team is focused on building new services and modernizing old ones, not just maintaining legacy code, which is where Copilot tends to earn its keep by helping engineers navigate and refactor complex older codebases.

For a company whose software quietly keeps goods and information moving around the world, having AI assistance baked into the development process is a meaningful shift in how that infrastructure gets built.

---

Mimecast

Cybersecurity · Remote · GitHub Copilot

Mimecast is one of those companies most people have never heard of but probably rely on every day without knowing it. They're a cybersecurity firm protecting more than 42,000 organizations worldwide from phishing, ransomware, email compromise, and the newer breed of AI-powered attacks. On a typical day, their systems analyze more than 18 billion security events.

What makes Mimecast interesting is that they've decided to treat AI-assisted development not as a perk but as a strategic function. The company is building out an entire internal team dedicated to developer experience, with GitHub Copilot at the center of how they plan to transform engineering across the organization.

The mandate is pretty concrete. Mimecast wants to measure the productivity gains from Copilot using real engineering metrics like cycle time and deployment frequency, establish best practices for how engineers use AI in their daily work, and make sure adoption is consistent across all their engineering teams.

They're candid that this is a cultural shift, not just a tooling one. The goal is to move engineers from AI-assisted to AI-native working patterns, treating Copilot as embedded in how teams work rather than an optional add-on.

For a cybersecurity company where code quality and reliability directly affect customer safety, the decision to centralize Copilot adoption across the engineering organization is a notable vote of confidence in the tool.

---

BlaBlaCar

Technology · Paris, France · GitHub Copilot

BlaBlaCar is the carpooling app that connects drivers with empty seats to passengers heading in the same direction, helping them split the cost of the trip. It's the world's leading community-based travel app, with 27 million members using it across 21 countries every year.

Behind that simple premise is a serious engineering operation. BlaBlaCar runs on Google Cloud Platform and Kubernetes, with backend services in Java, infrastructure tooling in Go, and data pipelines in Python. Keeping all of that running smoothly for millions of users across Europe and beyond takes a dedicated platform team.

That team, which they call Engineering Experience, is where GitHub Copilot fits in. Their job is to build and maintain the internal tools and infrastructure that let every other engineer at BlaBlaCar ship code quickly and safely. Copilot is part of their standard toolkit, alongside GitHub Actions, Jenkins, and Terraform.

What's interesting is the framing BlaBlaCar uses. They describe their approach as wanting to reduce toil and eliminate repetitive work by leveraging AI code generators and assistants. The goal is to make the lives of their developers easier so they can focus on the harder, more creative problems.

For a company that has helped avoid 2.5 million tonnes of CO2 emissions by getting people to share rides, having AI quietly speed up the engineering work behind the scenes fits naturally with a culture built around efficiency.

---

Zelis

Healthcare Technology · Morristown, New Jersey · GitHub Copilot

Zelis is one of those companies operating quietly in the background of the American healthcare system. They're a technology company that handles the financial side of healthcare, serving more than 750 payers including all top five national health plans, along with millions of providers and consumers.

With more than 2,000 employees and offices across the US and India, Zelis has reached the scale where how their engineers work day to day has a direct impact on how quickly new features ship to customers.

That's where GitHub Copilot comes in. Zelis is rolling out Copilot as a central piece of their company-wide AI strategy, treating it not as an optional perk but as part of the standard workflow for every software engineer. The goal is to drive consistent adoption across teams, build internal training programs to help engineers use it effectively, and measure productivity gains with real engineering metrics.

What's notable about Zelis is the cross-functional angle. Alongside Copilot for engineers, they're deploying Microsoft Copilot across the broader business and building custom AI agents for specific roles, with the engineering rollout being the anchor initiative. The company is also building dashboards to track adoption patterns and cost avoidance so leadership can see where AI is actually moving the needle.

For a healthcare financial technology company operating in a highly regulated environment, choosing to centralize AI-assisted development under a dedicated program is a sign that they see Copilot as a long-term strategic investment rather than a short-term experiment.

---