Decagon CEO Questions Enterprise AI's Reliance on Engineers

The Forward-Deployed Engineer Debate

As enterprise AI vendors race to embed technical staff inside client organizations—with Microsoft committing $2.5 billion and roughly 6,000 specialists to such programs, and AWS pledging $1 billion—Decagon CEO Jesse Zhang is making a contrarian argument: the need for these engineers proves the software is poorly designed, not that deployment inherently requires them.

Monthly job listings for forward-deployed AI roles surged more than 800% between January and September 2025, according to PYMNTS. Among executives at companies with at least $1 billion in annual revenue, 71% cite organizational readiness rather than technology as the primary barrier to AI performance, PYMNTS Intelligence found. Only 11% blame the models themselves.

Zhang told Newcomer his position stems from observing a single customer's experience. That client spent a year working with Sierra's forward-deployed engineers—Sierra being the customer-service AI company led by Bret Taylor that has raised three times Decagon's funding and reached $200 million in annualized revenue, according to equity research firm Sacra. During that year, the customer built three workflows. Zhang described the arrangement as a black box where any modification required going back through the engineers, who were eventually reassigned to other accounts.

Why it matters

The economics of enterprise AI deployment hinge on whether vendors can grow revenue without proportionally expanding implementation teams. If AI systems remain too complex for client staff to configure independently, the cost structure of the industry locks in permanent reliance on expensive technical labor—limiting both vendor margins and customer flexibility.

Decagon's Self-Service Bet

After switching to Decagon, Zhang said, the same customer built seven new workflows within about a month. He credits a product designed for clients' own staff, including non-technical employees, to operate directly.

Decagon does employ its own forward-deployed engineers, who get deployments live in roughly six weeks. The company grew revenue to about $35 million by October 2025 and added more than 100 enterprise customers, according to Sacra. Zhang's distinction is what those engineers leave behind: a product configured for client autonomy that eventually breaks the link between adding customers and adding implementation headcount. Roughly 90% of Decagon's workloads now run on fine-tuned open-source models rather than frontier models, which the company says are faster and cheaper for narrow, repeatable customer-service tasks.

Salesforce's Scale Play

Salesforce is testing a similar thesis at far larger scale. Its Agentforce product reached $1.2 billion in annualized recurring revenue in the first fiscal quarter of 2027, up 205% year over year. In June, the company agreed to acquire Fin, the AI agent business formerly known as Intercom, for about $3.6 billion.

Those figures aren't directly comparable to Decagon's growth. Salesforce's installed base and sales infrastructure provide distribution advantages that measure reach as much as product pull. The numbers represent different scales of the same market question: can enterprise AI become self-service fast enough to change its labor economics?

If buyers can eventually configure, expand and manage AI systems themselves, vendors can scale revenue without proportionally expanding implementation teams. If they cannot, the forward-deployed engineer remains a permanent cost, and the software that required one becomes harder to justify.

These details were first reported by PYMNTS.