---
title: AI Startup Autoheal Raises $7.9M to Clear the Incidents and Security Fixes AI Coding Leaves Behind
description: AI startup Autoheal raises $7.9M led by Innovation Endeavors for agents that handle incidents, security fixes and AI coding costs inside enterprise clouds.
author: Darie Nani (Editor-in-Chief)
date: 2026-09-28T12:28:49.502Z
updated: 2026-09-28T12:28:49.514Z
canonical: https://www.sovereignmagazine.com/article/autoheal-7-9m-seed-software-factory-ai-coding
image: https://cdn.nanimediahouse.com/autoheal-founders-ohm-choudhury-saraswat.webp
categories: Artificial Intelligence, Startups
content_type: Spotlight
region: San Francisco
publication: Sovereign Magazine
about:
  - type: Organization
    name: Autoheal
    description: Autoheal is an enterprise AI company whose agents handle the engineering work that follows the writing of code, including incident response, vulnerability remediation and the cost of AI coding tools. Its platform runs inside a customer's own cloud, connects to the coding agents and engineering tools a team already uses, and uses Evaluator and Healer agents to score and improve the other agents over time. It was founded by Sid Choudhury, Utkarsh Ohm and Puneet Saraswat and is backed by Innovation Endeavors, Emergent Ventures, U&I Ventures, Darkmode Ventures, Batch Ventures and Param Hansa Values.
    url: https://autoheal.ai/
    industry: Enterprise AI software
    sameAs:
      - https://www.linkedin.com/company/autohealai
      - https://x.com/autohealai
---

Autoheal, a startup whose AI agents handle the engineering work that follows the writing of code, from production incidents and security vulnerabilities to the cost of running AI coding tools, has raised $7.9 million in seed funding led by Innovation Endeavors, the venture firm co-founded by former Google chief executive Eric Schmidt.

Harpinder Singh, a general partner at Innovation Endeavors, joins Autoheal's board. Emergent Ventures, U&I Ventures, Darkmode Ventures, Batch Ventures and Param Hansa Values, the private investment firm of Nutanix co-founder Dheeraj Pandey, also invested. Autoheal will use the money to expand its platform, which it calls a self-improving software factory. Platform engineering teams at large companies use it to build, run and improve AI agents across the software development lifecycle.

Autoheal says Japanese bank Nomura has used its agents to cut its mean time to resolution, the average time taken to fix a production incident, from two hours to 15 minutes. AvidXchange, which sells accounts payable and payments software to mid-sized businesses, saves thousands of engineering hours a month and has moved that time to building new features, according to Autoheal.

## AI Coding Agents and the Rise in Engineering Toil

Google's 2025 DORA report found that 90% of software professionals now use AI in their work. In Veracode's 2025 GenAI Code Security Report, 45% of the code samples produced by more than 100 AI models [failed security tests](https://www.sovereignmagazine.com/article/anthropic-claude-desktop-browser-permissions), introducing flaws from the OWASP Top 10 list of the most common web application vulnerabilities.

In [Catchpoint's SRE Report 2026](https://www.catchpoint.com/blog/sre-report-2026-ai-optimism-and-the-economics-of-effort), the median share of working time that reliability engineers spend on toil, the repetitive manual work of keeping production systems running, rose to 34% from 20% a year earlier, the first increase in five years. Respondents said AI had created new tasks of its own, including maintaining AI tools, reviewing their suggestions and checking actions taken with AI.

Autoheal says repetitive work across the software development lifecycle (SDLC), such as incident response and vulnerability remediation, takes up more than a third of an engineering team's capacity, and that [controlling LLM spending](https://www.sovereignmagazine.com/article/microsoft-ai-token-budget-copilot-engineers) and managing the context coding agents need are now part of the same workload. Many large companies have platform engineering teams that run shared tools for their developers, and Autoheal says these teams are moving to a software factory model built on specialized AI agents. According to Autoheal, these rollouts often fail because the tools are fragmented, the agents do not share context and security rules restrict what they can reach.

## Autoheal's Software Factory and Its Evaluator and Healer Agents

> "Our experience taught us that while building the first version of an AI agent is easy, scaling it consistently across the enterprise SDLC is the real challenge."
> — Sid Choudhury

"Platform engineers need a unified platform to deploy agents that don't just execute tasks, but continuously improve alongside complex enterprise workflows," said Sid Choudhury, Autoheal's co-founder and chief executive. "We built Autoheal so they can immediately step into the role of AI engineers and accelerate ROI, without spending a year building the underlying infrastructure."

Autoheal connects to the tools an engineering team already uses, including coding agents such as Claude Code and [Codex](https://www.sovereignmagazine.com/article/openai-codex-plugins-ai-coding-tools), code repositories, CI/CD pipelines, observability tools, cloud environments and issue trackers. From these it builds an engineering context graph, a map of the company's services, infrastructure, runbooks and past incidents that all of its agents share. Worker agents handle specific jobs, among them incident response, vulnerability remediation, code review, release checks, support escalations and cost control.

Two more agents, the Evaluator and the Healer, run in the background. The Evaluator scores every run by a worker agent. For a coding agent, it scores the specifications and pull requests the agent produced against what happened next, including review comments, failed CI builds and any incidents the code caused. The Healer takes agents with low scores and opens pull requests that change their skills, prompts, tools or choice of model, then tests those changes against past benchmarks to check for regressions before an engineer reviews them.

Every change to an agent's behavior is stored in git and needs an engineer's approval, and the platform records what each agent accessed, its reasoning and its cost. Autoheal can run in a customer's own cloud, on its own servers or in an air-gapped environment.

Several startups sell AI agents for site reliability engineering (SRE), including Resolve AI and Traversal, which investigate production incidents. Autoheal also covers security fixes and coding costs, and works with the coding agents a team already uses.

## Incident Response Automation at Nomura and AvidXchange

"Our production operations teams spend valuable time triaging alerts and managing incidents, while also pulling engineers away from their software development activities," said Sameer Jain, chief information officer for wholesale at Nomura. "Autoheal gives us a platform that takes investigation timelines down from hours to minutes. The fact that it runs entirely within our own cloud, in compliance with our controls, made it a natural fit for how we operate."

"In production incident response, Autoheal took our time to root cause to minutes, with evidence our engineers trust. That's time our developers stay focused on feature work," said Krish Shetty, chief technology officer at AvidXchange. "Next, we're shifting it left into other critical parts of our SDLC, because every engineering hour we get back goes into shipping faster for our customers."

## Autoheal Founders From Harness, ThoughtSpot and Microsoft

Choudhury co-founded Autoheal with Utkarsh Ohm, its chief technology officer, and Puneet Saraswat, its chief development officer. Choudhury was previously senior vice president and general manager at Harness, the software delivery platform, and at database company Yugabyte, and was the first product manager at AppDynamics. Ohm led AI and machine learning engineering at analytics company ThoughtSpot. Saraswat was vice president of engineering at Harness after working as a principal software engineering manager at Microsoft.

The founders say that after helping grow Harness past $200 million in annual recurring revenue, they found that deploying AI agents safely across many engineering teams took far longer and used far more tokens than building a single agent. They concluded that a platform had to manage agents as code, with agents that learn from each run overseeing the others, so that companies do not end up with large numbers of unmanaged agents.

"Enterprises are moving quickly from experimenting with AI agents to asking how they can operate them safely and efficiently at scale across the entire software factory," said Singh, who co-founded data company Slice before its sale to Rakuten. "Autoheal is building the agent infrastructure layer that makes that possible. The opportunity is much larger than one agent or one workflow."

## Enterprise Small Language Models and Plans Beyond Software Engineering

Autoheal plans to use reinforcement learning on each customer's private engineering data to train small language models specific to that company, inside the customer's approved security boundary. These models would run the company's software factory agents at lower cost. Autoheal says frontier models know a great deal about the outside world but little about how a particular company works, and a model trained on the company's own engineering data is meant to address that.

In the longer term, Autoheal plans to apply the same system to data engineering and security engineering. The company expects every large enterprise to run a software factory with its own set of specialized agents, and wants to be the platform that engineering teams use to build, govern and improve them.

**About Autoheal**

Autoheal is an enterprise AI company whose agents handle the engineering work that follows the writing of code, including incident response, vulnerability remediation and the cost of AI coding tools. Its platform runs inside a customer's own cloud, connects to the coding agents and engineering tools a team already uses, and uses Evaluator and Healer agents to score and improve the other agents over time. It was founded by Sid Choudhury, Utkarsh Ohm and Puneet Saraswat and is backed by Innovation Endeavors, Emergent Ventures, U&I Ventures, Darkmode Ventures, Batch Ventures and Param Hansa Values.

[Website](https://autoheal.ai/)

## FAQ

**Q: What is an AI SRE?**
An AI SRE is software that does part of the work of a site reliability engineer, such as triaging alerts, investigating production incidents and finding their root cause. Autoheal runs an incident response agent of this kind and says it cut Nomura's mean time to resolution from two hours to 15 minutes. Other startups selling AI SRE tools include Resolve AI and Traversal.

**Q: What is the software factory in AI?**
A software factory is a way of running software development in which specialized AI agents handle repeatable jobs across the development lifecycle, from writing and reviewing code to fixing vulnerabilities and responding to incidents. Autoheal's version connects a company's coding agents, code repositories, CI/CD pipelines and observability tools, and gives every agent the same engineering context.

**Q: What are self-improving AI agents?**
Self-improving AI agents are agents whose instructions, tools or models are updated according to how well they performed. At Autoheal, the Evaluator agent scores each run and the Healer agent proposes fixes for low-scoring agents as pull requests, which are tested against past benchmarks and approved by an engineer before they take effect.

**Q: How can AI be used in incident response?**
AI agents can sort alerts, gather logs and metrics from different systems and point to the likely cause of a production incident, so engineers spend less time investigating. Sameer Jain, Nomura's chief information officer for wholesale, says Autoheal has taken the bank's investigation times from hours to minutes while running inside its own cloud.