The story of Kevin
A close Delaware family and a generalist's path
Kevin grew up in Dover, Delaware, as the middle of three siblings born within four years. His mother left a small Pennsylvania coal town, moved east, and became a teacher; his father grew up in Wilmington. Kevin describes a close household with meaningful opportunity and is careful not to claim his parents' hardships as his own. At 18, he had no elaborate university plan.
Strong mathematics scores led him into engineering at West Virginia University, where systems engineering fit his preference for a generalist view across technology and business. The family emphasis on relationships remained important. Years later, Kevin recorded a cross-country drive with his father as a long interview, learning family stories he had never heard and turning travel into a deliberate act of connection and preservation.
Imposter syndrome and co-founder lessons
Kevin began his career in Deloitte's systems integration practice. Surrounded by colleagues from schools he had never considered applying to, he experienced sustained imposter syndrome even as he built consulting, technology, and client skills. Later, he and a partner bootstrapped a company from the partner's parents' basement in Boston.
They believed their abilities were complementary, but over time the relationship operated too much like a co-CEO structure. The venture did not become a lasting business, yet it gave Kevin practical experience in selling, product development, partnerships, and technical delivery while the personal cost of failure was still manageable. He now sees a clearer division between his own business and product strengths and the deep architecture skills a technical co-founder should bring, with shared vision providing the essential overlap.
A marketplace for custom AI engineering
Agent Graph addresses organizations that know AI matters but do not yet know which use cases deserve investment or how to build them. A business can develop a custom agent specification through an assisted workflow, then send it to a marketplace where qualified AI engineers can bid on implementation. The customer owns the resulting source code rather than being locked into a generic vertical software product.
Kevin Denman says production agents need a method and infrastructure for agent-specific evaluation and benchmarking because their behavior is nondeterministic. He argues that this work belongs early in the design cycle, rather than being added after a prototype, so the team can assess behavior and improve the agent over time. Listen at 1:35:52
Agent Graph also provides forward deployed engineers who work inside an organization long enough to identify valuable opportunities before prescribing a solution. Kevin sees this translation layer as essential because leadership teams often have more urgency than technical clarity. His aim is to connect enterprise problems with engineers who can design, test, and deliver custom agents while meeting each organization at its current level of readiness.