I've been an SRE for 15+ years. Currently managing observability for 600+ APIs handling 200B+ daily transactions.
The problem that's haunted me the whole time: your infrastructure already knows why it's broken at 3am. The answer is in your logs, metrics, and traces. You just can't ask it. Instead you're hunting through dashboards, writing PromQL, correlating timestamps manually.
So I built ReductrAI. It's a proxy that:
Sits between your infra and your existing observability tools
Compresses logs/metrics/traces ~80%-99%, stores locally
Forwards everything to Datadog/Splunk/whatever unchanged
Builds an AI layer across all your data
Lets you query in plain English: "Why did latency spike at 2am?" "What changed before checkout started failing?"
Your data never leaves your environment. No migration. Setup is changing one endpoint.
Works with OTEL, Prometheus, Datadog, Splunk, syslog - 30+ formats. On-prem, cloud, Kubernetes, mainframe.
Still early. Running in production. Looking for feedback from anyone who's felt this pain.
hermosillo_21•56m ago
Sits between your infra and your existing observability tools Compresses logs/metrics/traces ~80%-99%, stores locally Forwards everything to Datadog/Splunk/whatever unchanged Builds an AI layer across all your data Lets you query in plain English: "Why did latency spike at 2am?" "What changed before checkout started failing?"
Your data never leaves your environment. No migration. Setup is changing one endpoint. Works with OTEL, Prometheus, Datadog, Splunk, syslog - 30+ formats. On-prem, cloud, Kubernetes, mainframe. Still early. Running in production. Looking for feedback from anyone who's felt this pain.