Wing's Tanay Jaipuria: "DeployCos" built on forward deployed engineers are becoming how enterprise AI gets shipped

Tanay Jaipuria, a partner at Wing, used his October 13 newsletter to name a pattern FDEs have been living inside all year: the rise of "DeployCos", companies whose job is to embed engineers inside enterprises and get AI agents running in real workflows. He points to Anthropic's $1.5B services venture with Blackstone, Hellman & Friedman and Goldman Sachs (launched in July as Ode with Anthropic), OpenAI's TPG-led deployment company that raised more than $4B and started with 100+ FDEs via its Tomoro acquisition, and AWS's $1B forward deployed engineering unit, alongside longtime practitioners Palantir and Accenture.

His explanation for the boom is a diffusion gap: models can already do far more than most organizations use them for, and closing that gap takes system access, workflow redesign and organizational buy-in that no model handles on its own. For the labs, he adds, deployment work also produces long-horizon traces of real work, creates lock-in when model quality is close, and lets vendors charge for outcomes rather than tokens.

Jaipuria sorts the new players into three groups: horizontal DeployCos such as the labs' ventures, Distyl and BrainCo; function-specific ones such as Sierra and Decagon in customer support; and vertical ones such as Tennr in healthcare and HappyRobot in logistics. He also argues that many vertical AI app companies selling to large enterprises are DeployCos in practice, because custom integrations, evals and change management make heavy FDE work unavoidable.

For FDEs, the essay describes a common playbook: start with a few bounded, checkable workflows, build the agent harness and evals, deploy once quality clears a bar, then add a learning loop in production. It also suggests FDE demand is spreading well beyond the labs, to any vendor whose product needs last-mile work inside each customer.