Herald

What it does

Herald is an AI SRE for engineering teams. It builds a context graph from observability data, code, CI/CD pipelines, docs and dependencies, then detects anomalies and investigates possible root causes across code, infrastructure and telemetry. It also triages alerts and answers questions about code, infrastructure, past incidents and runbooks. The site offers a free trial, demo booking, pricing, documentation and a contact form.

Herald works on both sides of the incident lifecycle: proactive detection and reactive help during incidents. The documentation lists alert triage, predictive incident detection and developer Q&A as use cases, with integrations that include AppDynamics, AWS CloudWatch, Datadog, Dynatrace, GitHub, Grafana, Honeycomb, Jira, Linear, New Relic, Opsgenie, PagerDuty, Sentry, Slack, Splunk and Zendesk. Agents can be deployed to Slack and Microsoft Teams, and there is a Python SDK, a CLI and release notes. The docs state the platform is SOC 2 Type II, read-only by default with approval-gated actions. A contact form asks for name, work email and monthly incident volume. The site notes Herald was previously known as RunLLM, and the footer still reads RunLLM, Inc.