How Coveo cut debugging time by 50% across 1,200+ monthly releases

Coveo helps businesses use AI and GenAI to deliver search, recommendations and generative experiences across workplace, website, customer service and commerce use cases. Their technology makes it easier for employees and customers to find relevant content quickly, whether on a website, on an ecommerce site, or within internal systems. By personalizing search results, Coveo ensures users get accurate, relevant answers, boosting proficiency and enhancing the overall experience.

—------

Supporting large global brands with millions of end users relying on their systems, ensuring a secure, reliable, and bug-free experience isn’t just important—it’s critical. But as Coveo scaled to over 230 engineers shipping platform updates on average 50 times a day, their self-hosted solution started to show cracks.

Fragmented observability across 1,200+ releases monthly

To meet strict security requirements, Coveo initially ran Sentry on-premises. While this worked at first, the setup—separate tenants for each team, region, and environment—became more complex to manage. As the company grew, so did the challenges of maintaining visibility and consistency across their systems:

• Disconnected environments: Separate tenants - while a stop gap like solution to account for security requirements - created duplicate alerts and missing context, making it harder for teams to identify and resolve issues quickly.
• Limited production visibility: Infrastructure and maintenance costs restricted certain teams from accessing observability in production, leaving them to rely on customer reports to identify bugs.
• High maintenance costs: Managing multiple on-prem instances required significant resources to handle updates, governance, and security, taking time away from other priorities.
• Growing complexity: With a diverse tech stack spanning Java, Python, Go, Scala, and more, maintaining consistency across fragmented systems became challenging.

Every region had its own environment, and tracking issues across them was complex. We’d get the same error reported in multiple places with no central visibility

Kevin Larose, Senior Technical Product Manager at Coveo.

Requirements

To address these challenges, Coveo required a solution that could provide clarity, reduce operational overhead, and give teams the context they need to debug quickly and proactively address issues before they impacted customers.

• Agnostic platform for centralized error tracking: Consolidate observability across 100 teams in Java, Python, JavaScript, Node, and more into one platform, eliminating the inefficiencies and blind spots created by separate tenants for each environment.
• Debugging tools with context: Enable teams to detect, contextualize, and resolve issues before customers reported them, improving reliability and customer satisfaction.
• Governance and Security: Meet rigorous data governance requirements handling PII to ensure compliance with data residency regulations, and maintain high standards of security.
• Integration with existing tools: Connect seamlessly with tools like GitHub, JIRA, and Slack to streamline workflows and reduce debugging time.

As a platform engineering team, our goal is to enable others to succeed. Moving to SaaS removed blockers for many teams and ensured no one was flying blind anymore

Kevin Larose, Senior Technical Product Manager at Coveo.

Moving to Sentry SaaS

Coveo ultimately chose to move to Sentry SaaS, as it provided the code-level visibility and debugging context they needed across their distributed services, all while meeting strict security requirements and eliminating the burden of maintaining a complex on-prem infrastructure. Sentry’s SaaS offering provided the team with the stability, centralization, and flexibility they had been missing. For Coveo, the decision to move to Sentry SaaS was fueled by:

• The latest features: Access to Sentry’s full suite of features, including real-time error tracking, session replay, and performance monitoring, deep integrations with GitHub and Slack – all without the need to maintain infrastructure.
• Seamless scaling for new teams: Centralizing error tracking across environments enabled other teams within Coveo to adopt Sentry easily, supporting their rapid growth and making it easy for teams to onboard and gain visibility into production.
• Enhanced security and compliance: With Sentry’s robust data governance capabilities, including the use of Relay —a powerful tool for controlling what data is sent to Sentry—Coveo could meet their security and compliance needs without sacrificing innovation.

“We’re shipping six to seven times a day across more than 100 teams,” said Kevin Larose, Senior Technical Product Manager at Coveo. “Keeping up this pace with the visibility we had was nearly impossible before moving to Sentry SaaS.”

Migrating 250 projects in less than 2 months

Migrating 250 projects and 230 engineers to Sentry SaaS took less than two months - faster than expected. Teams that previously lacked visibility into production issues now had real-time insights, cutting troubleshooting time from days to hours— and sometimes minutes.

As Germain, platform lead at Coveo put it, “The visibility provided by Sentry is crucial in maintaining a stable platform, allowing the team to ship updates faster and with greater confidence.”

An example was when one team at Coveo accidentally swapped a GET request for a POST in their code, causing errors to surface immediately after deployment. Before Sentry, finding and fixing such issues meant hours of digging through logs and manually tracing the root cause.

Sentry flagged the error as soon as it occurred and with the GitHub integration identified the suspect commit, adding a comment directly in the PR and notifying the responsible developer. Within minutes, the developer created a follow-up PR to fix the problem, and the corrected code was deployed in record time. This deeply integrated workflow not only reduced debugging time, but gave Coveo the confidence to ship faster without worrying about undetected errors in production.

A major win for us was the integration with Slack and GitHub. I wrote one bad line of code and it made it to production, Sentry immediately flagged it. It sent me a Slack notification and a GitHub alert, pointing straight to the exact line of code where I messed up. Shaving hours off the debugging process,” said Jean-Phillipe, Staff Software Developer at Coveo.“A major win for us was the integration with Slack and GitHub. I wrote one bad line of code and it made it to production, Sentry immediately flagged it. It sent me a Slack notification and a GitHub alert, pointing straight to the exact line of code where I messed up. Shaving hours off the debugging process

Jean-Phillipe, Staff Software Developer at Coveo.

Final Thoughts

For any team considering a move to Sentry.io, the advice is simple: “Just start using it. You’ll see the difference.”

Since implementing Sentry SaaS, Coveo has transformed its development workflow. Their ability to detect, prioritize, and resolve issues before they impact customers has dramatically improved:

• 50% Decrease in Debugging Time: Stack traces and Slack alerts enable engineers to respond immediately.
• Increased Release Velocity: With confidence in Sentry’s monitoring, the team doubled their release pace without risking undetected issues in production.