"We investigate every Sentry alert — no exceptions. Struct made that sustainable by eliminating manual triage. Senior engineers got their time back for higher-leverage work, and the broader team adopted it without friction."

Siva Sooriyan

CTO, Arcana

About Arcana

Arcana is a fintech company that enables institutional investors to understand portfolio risks, decompose single stock & book performance, drill into crowding, and isolate idiosyncratic differentiation. Their customers have trillions of AUM.

The Challenge

Arcana's team was getting hundreds of high priority Sentry alerts monthly, and every single one of them had to be investigated and triaged.

Every morning, the team spent hours cleaning up every single exception, with individual investigations averaging 10-45 minutes each (longer for major incidents) - 15+ hours per week on Sentry triage alone to ensure that they were delivering the most reliable software possible. Newer engineers couldn't meaningfully contribute because effective triage required deep institutional and codebase knowledge.

The Solution

Arcana integrated Struct with Sentry, GitHub, GCP Cloud Logging, and Slack. When a Sentry exception fires, Struct auto-investigates, querying logs, analyzing stack traces, reviewing recent commits, and more. By the time an engineer opened Slack to look at the alert, Struct had posted a full investigation with customer impact, root cause, and pattern analysis. To test alternative hypotheses or dive deeper, they simply tagged Struct in a reply directly in the thread.

Setup was straightforward: they connected their observability stack, pointed Struct at their Slack channels, and investigations started running automatically.

The Result

Triage time dropped from hours to minutes per day. The on-call engineer now reviews Struct's investigation output instead of manually querying logs and cross-referencing code. As confidence in Struct grew, investigation coverage scaled 17x while maintaining an >80% helpful investigation rate.

"I often skip Sentry entirely now — I just read the Struct output and pass it to the relevant person. Morning exception cleanup went from taking me an hour to about 10 minutes. And it helps our newer engineers triage issues they wouldn't have been able to handle on their own before."

Yash Sahijwani

Software Engineer, Arcana

The team's senior engineers got their time back to focus on higher-value work instead of spending hours a day on alert triage. A newer engineer was also successfully onboarded onto on-call, leaning on Struct's investigations to jumpstart understanding and triage. The broader team has also started using and referring to Struct investigations, with # of users organically 10’xing after the pilot.