FDE: A Platform Instead of Custom Development (Category AI4SDLC)

I watched Kevin Bai's 18-minute talk, “Forward Deployed Engineering 101,” which continues the FDE theme. In the previous review, I was interested in what such an engineer does hands-on; here, the question is why the business needs the role at all and how to avoid turning implementation into expensive custom development. Kevin is now on Anthropic's Applied AI team; previously, he built the FDE team at Rippling and worked at Palantir. He therefore explains the model not as a fashionable role but as a way to sell a complex platform.

The framework is simple: an FDE is needed when a company sells a technically complex system to a nontechnical buyer. CTOs and developers usually learn a platform themselves; a nontechnical team only needs Jira or Slack to be configured. The difficult quadrant is a platform that must be built upon and a customer without deep engineering capability of its own.

A license alone is not enough. An FDE is a developer who can be trusted to talk with the customer. The engineer understands the real process and assembles a solution on top of the platform. The customer buys neither a box nor developer hours, but a working outcome. There is an important boundary, however. If every FDE writes a separate system from scratch for each customer, this has become a custom development studio, and the economics will drown in support costs. Shared building blocks are essential: what is unique stays with the customer, while what repeats goes back into the platform. Implementation then also becomes a channel for product discovery.

AI makes assembly cheaper but does not repeal this rule. Kevin's hypothesis is that agentic platforms are easier to adapt, so more companies may need the FDE model. I would add that cheap code increases the risk of proliferating disposable solutions. The faster the implementation, the more important architectural discipline becomes.

My conclusion is that two questions must be answered before hiring FDEs. Are we really selling a complex system to a nontechnical buyer? And is there a platform on which solutions can be assembled repeatedly? If the answer to the second question is no, the fashionable role name will merely disguise custom development inside a product company.

#AI #AI4SDLC #Engineering #Architecture #Product #Management