Dodge.ai builds autonomous agents for enterprise ERP systems, tackling a problem that's historically been solved by throwing more consultants at it: keeping SAP running. Where legacy maintenance work has meant endless billable hours for outside consultants, Dodge is betting that agents — not people — should be doing that grind instead.

"For decades, keeping SAP running meant throwing consultants at it. Dodge AI throws agents at it instead — and the agents don't bill by the hour."

Cutting through the wrapper problem

Dodge.ai's traction is notable in context: when Google and Accel reviewed more than 4,000 applications for their India-focused Atoms accelerator, roughly 70% were dismissed as "wrappers" — AI features layered on top of existing software without reimagining the underlying workflow. Dodge was one of only five startups selected for the cohort, alongside K-Dense, Persistence Labs, Zingroll, and LevelPlane, earning up to $2 million in funding and up to $350,000 in Google cloud and AI compute credits.

From an India-only story to US enterprise traction

Before Embark, Dodge had zero presence with US enterprises — the SAP maintenance problem it was solving was still an India-only story. Since then, the company has landed 3 US enterprise customers with 7 more in active pilots, and raised $2M led by Accel India directly off the back of that momentum.

From Rebhav, Co-founder, Dodge AI: "This experience has been a real eye-opener, giving me the opening I needed to venture into the US market finally. Before Embark, we had zero presence with US enterprises — the SAP maintenance problem we were solving was still an India-only story. Since then, we've landed 3 US enterprise customers and 7 more in active pilots, and raised $2M led by Accel India and Google AI Futures Fund directly off the back of that momentum."

With a spot in one of the most selective AI cohorts around, early US enterprise traction, and fresh capital in the bank, Dodge is betting that agents — not consultants — are what modern SAP maintenance should look like.

‍

‍