A San Mateo startup founded by former Harness and ThoughtSpot executives is betting that companies will pay for software that watches their AI systems and fixes problems before engineers notice them. Autoheal said on September 28, 2026, it raised $7.9 million in a seed round led by Innovation Endeavors, with backing from Emergent Ventures, U&I Ventures, Darkmode Ventures, Batch Ventures' CTO Fund and Param Hansa Values.
The company declined to share its valuation. Founded in 2025, Autoheal builds what it calls autonomous agents that monitor AI deployments in production and repair issues on the fly.
The round drew checks from a handful of enterprise software veterans: Shawn Kung, who founded GIT100; Sumeet Arora, chief product officer at Teradata; Anshu Sharma, co-founder and CEO of Skyflow; Savin Goyal, co-founder and CTO of Outerbounds; and Srikant Gokulnatha, a former senior vice president at ThoughtSpot.
Autoheal's three co-founders spent years building infrastructure software at scale before launching the company. CEO Sid Choudhury served as senior vice president and general manager overseeing DevSecOps and developer experience products at Harness, and held product leadership roles at Yugabyte, AppDynamics, Salesforce and Oracle. CTO Utkarsh Ohm ran AI and machine learning engineering at ThoughtSpot and was among the first engineers at HyperTrack. Chief data officer Puneet Saraswat was vice president of engineering at Harness and spent time as a principal software engineering manager at Microsoft.
How It Works
The platform Autoheal built is more elaborate than a simple monitoring tool. The company describes it as a "self-improving software factory" designed to handle incident response, patch security vulnerabilities and manage the costs of AI-generated code.
Two systems run underneath the hood. A Context Plane connects code repositories, microservices, ticketing systems and release pipelines into a single graph, complete with what the company calls a skills and memory registry. A Control Plane sits on top, enforcing spending limits, model selection rules, approval workflows and access policies.
Each time an agent runs, an "Evaluator" component scores its performance by tracking downstream signals such as code review feedback, how many times continuous integration pipelines need to re-run, new alerts and whether the root cause analysis turned out to be correct. A second component, called "Healer," then generates pull requests that propose updates to skills definitions, memory stores, the scope of tools the agent can use, or which AI models it should call. Engineers review and approve those changes, the company said.
The platform hooks into Datadog, GitHub, Sentry and other widely used developer tools through a command-line interface, Slack and Microsoft Teams. Autoheal said it supports bring-your-own-cloud setups and air-gapped deployments for customers with strict data residency or security requirements.
Early Customers
Two sizable enterprises have gone on the record as production users. Sameer Jain, wholesale chief information officer at Nomura, said in the company's announcement that operations teams at the Japanese bank had been spending hours triaging alerts and pulling engineers off development work to manage incidents. "Autoheal gives us a platform that takes investigation timelines down from hours to minutes," Jain said.
Krish Shetty, chief technology officer and senior vice president at AvidXchange, a payments automation company, said the platform "took our time to root cause to minutes, with evidence our engineers trust." A HackerNoon article mentioned that Empiric Earth and Oscilar also use Autoheal, though neither firm appears on the company's public customer list.
Autoheal claims the platform reduces mean time to resolution by up to 80 percent, speeds vulnerability remediation by a factor of three and cuts AI coding costs by as much as 40 percent. The company did not offer independent benchmarks or third-party audits to support those figures.
On compliance, Autoheal said it holds SOC 2 and ISO 27001 certifications. Saraswat wrote in a LinkedIn post in June that the company had cleared ISO/IEC 27001:2022 and SOC 2 Type 2 audits.
Crowded Field
The market for autonomous operations tools has gotten noisy in recent months. Datadog, the publicly traded observability giant, launched its Bits AI SRE Agent in December and has since expanded the Bits AI product line, a clear signal that incumbents see revenue potential in automating site reliability engineering work. Furl, a startup focused specifically on vulnerability remediation, raised $10 million in January.
Other entrants include Kestrel, Nova AI Ops, NeuBird and Hyperagentic, all pitching versions of agent platforms aimed at platform engineering or DevOps teams. A July report from Checkmarx found that 96 percent of developers now use AI coding tools, but only 18 percent apply security checks continuously as they write code. That gap represents the opening Autoheal and its competitors are chasing with automated remediation.
Harpinder Singh, a general partner at Innovation Endeavors who led the round, will join Autoheal's board, according to media reports, though the company has not officially confirmed the board appointment.
