Arga Secures $10 Million Seed Round to Simulate Enterprise Software for AI Agents

Arga, a startup focused on improving AI agent training, disclosed a $10 million seed round headed by General Catalyst and supported by Box Group, Emergence, Gradient and SV Angel. The firm constructs comprehensive digital twins of enterprise applications—including Salesforce, Workday and various email clients—preserving permission structures and webhook integrations. This approach offers a repeatable sandbox where agents can be tested and refined across multiple systems, addressing challenges such as recognizing that a lead in Salesforce and a contact in Hubspot refer to the same company.

CEO and co-founder Philip Li illustrated how the platform can verify whether an email has already been sent and determine the appropriate recipient among overlapping opportunities. Traditional reinforcement-learning methods struggle with enterprise software because resetting environments like Salesforce is impractical; Arga’s solution enables rapid scenario resets and parallel environment runs. General Catalyst’s managing director Yuri Sagalov emphasized that such testing infrastructure is increasingly vital as agents become central to business application usage. The funding will support further development of these simulation tools, potentially accelerating the adoption of AI agents across corporate workflows.