We built Salesgraph after living the problem.

Ricardo spent 3.5 years in sales before becoming a founding engineer at Mintlify. Ruhan worked at Microsoft and Amazon before becoming Mintlify's founding solutions engineer.

At Mintlify, enterprise deals slipped when the work around them outgrew our capacity. We built an internal tool that turned revenue context into the work each deal needed. We left to start Salesgraph and joined Y Combinator to solve the same problem for other teams.

After launch, Global 2000 teams asked how they could move revenue workflows from predominantly human-run to predominantly agent-run without replacing every system at once.

Answering it led us to build the governed execution layer behind Salesgraph. Agents use the company's revenue context to finish work. We are still building the tracing and controls teams need to see what each agent did and why.

Revenue processes look similar from a distance, but the details vary by rep, product, customer, and industry. Public data is especially sparse in enterprise. Agents need the company's own context and a way to learn from results.

We are working on two hard problems. Teams need a complete trace of what agents did and why. Agents need a context layer that gets them to the right evidence quickly and consistently.

Revenue workflows also need agents and training environments built for the job. Better models alone do not solve this. The work also depends on revenue judgment and knowledge learned inside the company.

We have worked on both sides of this problem, in sales and engineering, and built the first version at Mintlify.