AI site reliability engineering
Deductive AI was founded by Rakesh Kothari (CEO), an early ThoughtSpot engineer who led distributed query processing teams, and Sameer Agarwal (CTO), a UC Berkeley PhD who created BlinkDB and was among the first engineers at Databricks building Apache Spark. The company formally launched in November 2025 with a mission to end the 3 a.m. debugging scramble.
How it works
Deductive builds a knowledge graph that connects your codebase, telemetry (logs, metrics, traces), engineering discussions, and knowledge bases into one queryable system. When an incident hits, multiple coordinated AI agents test hypotheses against this graph, tracing symptoms back through services, commits, and infrastructure to root causes — with evidence attached. Teams can also ask natural-language questions like "why did latency increase?" The platform integrates with Datadog, Prometheus, Grafana, Splunk, New Relic, Elastic, OpenTelemetry, PagerDuty, Jira, and Slack, and offers both SaaS and self-hosted deployment with no training on customer data. DoorDash reported over 1,000 engineering hours saved; Foursquare and Kumo are also customers.
Funding
Deductive announced a $7.5M seed round in November 2025 led by CRV, with Databricks Ventures, Thomvest Ventures, and PrimeSet participating, claiming incident resolution time reductions of up to 90%.