Dodge AI Raises $2.65 Million to Automate Enterprise Software Maintenance

Dodge AI has raised $2.65 million to tackle a problem that rarely ends with one broken application: a failure can depend on rules scattered across a company’s software. The startup says its AI can trace and resolve those failures while recording what it learns. For now, its customer count and examples of faster repairs are company-reported.

A round for the maintenance layer

The September 29, 2026, round was led by Accel and Google’s venture arm. New Build Venture Capital, Antler, Schema Ventures and angels from the SAP ecosystem also participated. Dodge AI presented the financing alongside its plan to handle incidents and requests for changes across enterprise applications; it did not describe a separate product launch.

The work Dodge wants to take on is more than answering a support ticket. Its platform is designed to connect business processes, customized software, service-management records and older configurations to identify a failure’s cause and recommend a fix. The applications it names include SAP, Salesforce, Microsoft Dynamics, Kinaxis and Oracle JDE. A repair that spans those systems may require knowing why a company changed a standard rule years earlier.

The repair is also meant to leave a record

Dodge’s longer-term pitch is to preserve the exceptions and custom rules uncovered during maintenance, rather than let that knowledge remain scattered among tickets, configurations and people. It wants that record to guide AI agents working in live business systems. That makes the immediate repair and the proposed operating manual parts of the same bet: an agent needs the company-specific rules behind a problem, not just an answer to the latest ticket.

132×Inventory-planning speedup

Dodge AI says it made one customer’s inventory-planning process 132 times faster after changing a process that had been run overnight because SAP crashed when it ran in the morning.

What the customer examples show

Dodge says it is working with more than a dozen enterprises, about half of them publicly listed, and that its platform handles hundreds of queries each hour. Those figures indicate use, but the announcement does not identify the customers behind its repair examples. It offers two cases to show the kind of cross-system work the platform is meant to do.

In one, Dodge says a truck could not load because a warehouse document printed incorrect information. The platform traced the fault across SAP, Kinaxis and internal warehouse software and delivered a fix within minutes. In the inventory-planning case shown above, Dodge also says the change improved order-allocation time by eight hours. Both accounts come from the company; the announcement provides no independent assessment of the fixes or the speedup.

Accel investor Prayank Swaroop argues that maintenance can give Dodge the context AI agents need before they operate safely in production. The distinction to watch is whether the platform can make its record of custom rules dependable across customers, not just resolve individual incidents quickly. Dodge has described the ambition and supplied examples, but has not shown how broadly those results hold.