Sherlocks AI
Sherlocks AI provides an AI SRE platform that automates production incident investigation and root cause analysis. The platform deploys 16+ specialized AI agents that correlate data across existing observability stacks, with a read-only data agent running inside the customer's VPC. It integrates natively with Slack and Microsoft Teams, delivering investigation results with evidence directly in the communication channel where alerts arrive.
- Artificial Intelligence
- AI Agents
- Developer Tools
- Software Only
- Data & Analytics
Funding
Founders
Founder details are not available yet.
Product
Problem
When alerts fire in production, the first alert rarely names the actual cause, leaving on-call engineers to manually dig through logs, metrics, and traces to diagnose what went wrong. This investigation phase is the biggest bottleneck in incident response, often taking 40 minutes or more per incident and contributing to engineer burnout, especially for teams with noisy alerts and incomplete runbooks.
Solution
Sherlocks AI automates the investigation phase of incident response by receiving alerts from existing monitoring tools and dispatching AI agents to analyze the underlying telemetry data. The platform's Watson data agent runs inside the customer's VPC with read-only access, connecting directly to underlying systems (databases, Kubernetes, cloud providers) rather than waiting for the observability layer to explain them. Investigations complete without engineers leaving their communication tools, and all interactions are fully encrypted for end-to-end protection. The system supports multiple LLM providers (Anthropic, OpenAI, AWS Bedrock, open-weight models) to avoid vendor lock-in and maintains a persistent knowledge graph, learning from user corrections to improve future root cause analysis.
Target Audience
Primary customers are SRE and DevOps teams with on-call rotations who need to reduce time-to-root-cause for production incidents. The platform serves engineering teams at 50 to 500-person product companies, as well as regulated teams in finance, healthcare, and the public sector that need the agent to run inside their own network.
Features
- 16+ specialized AI agents that automatically investigate alerts by correlating data across logs, metrics, traces, deploy history, and cloud infrastructure
- Watson data agent runs inside the customer's VPC with read-only credentials, ensuring data never leaves the customer's controlled environment
- Multi-LLM support with tested failover across providers, ensuring investigations continue even when one AI vendor experiences outages
- Persistent knowledge graph that remembers service dependencies, past incidents, and user corrections made in Slack, so new on-call engineers inherit context
- Broad integration list covering Datadog, New Relic, Prometheus, Grafana, Kubernetes, AWS, GCP, Azure, Sentry, ELK, Loki, databases, queues, CI/CD, and version control
- SOC 2 Type 2 certified with options for air-gapped or in-VPC LLM deployments for regulated industries
- Slack and Microsoft Teams native workflow that delivers investigation results with full audit trails of every hypothesis tested
- Event submission caps at 250,000 events per minute per org, and the agent is read-only by default, never executing changes in production systems