We built Devloom because we've been that on-call engineer.
It's 2:17am. Your phone is going off. You have 12 alerts and zero idea which one is the cause. You open four tabs. You run the same grep six times. And an hour later, it was a connection pool config that shipped in the last deploy.
How it started.
From 2020 to 2024, Marcus Webb was platform lead at Vantapay, a B2B payments fintech in Seattle running a distributed system across 63 microservices. The product handled time-sensitive settlement flows where a 15-minute P1 cost real money. Every major incident meant the same ritual: open Datadog for metrics, switch to Grafana Loki for log queries, pull Jaeger for trace waterfalls, paste correlation IDs into Slack, repeat until something clicked.
"After the sixth time I found the root cause was a connection pool config change that shipped 20 minutes before the alert fired, I stopped accepting that this was just how incident response worked."
Marcus left Vantapay in early 2024 and spent six months building the causality layer he kept waiting for someone else to ship. Priya Nakashima joined from a distributed tracing role at a cloud provider in mid-2024. Jordan Kessel joined in late 2024 to own developer experience and integrations. Three people, Seattle, bootstrapped deliberately — a focused tool doesn't need a sales team.
We are not trying to replace Grafana or Datadog. Those tools are good at what they do. We answer the one question they don't: which specific deploy, config change, or upstream dependency caused this alert, and what exactly do I revert?
What we believe.
Root cause, not root noise.
Observability tools show you everything. We show you what matters. Devloom discards 94% of correlated alerts by default — not because we're hiding data, but because most of it isn't causal. This isn't a general-purpose observability platform. It's specifically for finding the cause and the fix. If you want prettier dashboards, there are better tools for that.
Your stack, not ours.
We don't ask you to rewrite your pipelines. If you run Prometheus, Loki, and Jaeger today, Devloom reads them. We connect in minutes, not months. No proprietary agents.
Human in the loop, always.
We suggest. You apply. Devloom will never automatically roll back a deployment or modify production configuration. The model proposes. The engineer decides. Always.
Honest about limitations.
If the model isn't confident, we say so. Every root cause candidate has a confidence score and an explanation chain. A 41% confidence answer displayed honestly is more useful than a 95% confidence answer you can't interrogate.
The team.
Marcus Webb
CEO & Founder
Four years as platform lead at Vantapay managing a 63-service payments platform. Built Devloom in 2024 after repeatedly finding P1 root causes in deploy diffs that should have been obvious from the start.
Priya Nakashima
Head of Engineering
Built distributed tracing infrastructure at a cloud provider and contributed to the OpenTelemetry collector. Knows the OTLP spec from the wire level up. Joined Devloom in mid-2024 to build the correlation engine.
Jordan Kessel
Developer Experience
Three years building CI/CD integrations and CLI tooling at a DevOps SaaS. Joined Devloom in late 2024 to own integrations and the CLI — specifically because the first-10-minutes experience in observability tooling is consistently terrible.
We're hiring.
Specifically: distributed systems engineers who've been on-call and hated it. If you've ever console-dived across four tools at 3am and thought "this is fixable," you might be the right person.