YC Startup Tracker

HyperProbe is pairing an AI on-call agent with read-only live probes that investigate suspected failures without restarting or redeploying services.

Backed by Y Combinator, the startup is targeting hard-to-reproduce issues while offering customer-infrastructure or private-VPC deployment and a free-to-paid service pricing model.

Tracking: Y Combinator

Geography: Mountain View, San Francisco Bay Area, United States

1. HyperProbe Launches AI On-Call Agent With Y Combinator Backing

HyperProbe has launched an AI on-call agent for engineering teams, with backing from Y Combinator.

The startup says its system connects alerts from PagerDuty, Datadog, or Slack to logs and traces, then places a virtual breakpoint on a suspected line while the service keeps running.

Its read-only probe captures live state without redeployment or restart, producing evidence for root-cause analysis.

The product targets failures that conventional telemetry or local reproduction can miss, including silent writes, race conditions, duplicate processing, and third-party contract changes. HyperProbe supports Node.

js, TypeScript, Java, Python, and Kotlin, and offers deployment in customer infrastructure or a private VPC; pricing starts free for one service, followed by $99 monthly per service, with the first investigation free.

Key facts:

- HyperProbe is backed by Y Combinator.

- Probes capture live variable state without redeployment or application restart.

- The product supports Node.js, TypeScript, Java, Python, and Kotlin.

- Professional pricing is $99 per service monthly, or $79 annually.

- The founders previously spent three years building production SDKs for a testing product.

Why it matters: HyperProbe is trying to move incident response beyond post hoc logs and traces toward evidence gathered from running production systems.

If the approach works as described, engineering teams could investigate failures that are difficult to reproduce without restarting services or deploying diagnostic code.

The read-only, non-blocking design, immutable audit trail, approval gates, PII redaction, and private deployment options address operational and data-governance concerns that could limit adoption.

The key signals ahead will be whether teams convert from the free investigation, how reliably the agent confirms root causes, and whether coding agents can use the workflow safely through its MCP server.

Dealroom