"FDEs aren't supermen": a Japanese founder on what really makes forward deployed engineering work

ITmedia Business Online has published a write-up of a CEATEC 2026 conference session on forward deployed engineering given by Daisuke Yamashita, CEO of Tokyo AI startup Privacy Tech. He defines an FDE as an engineer who solves problems on the customer's site, then folds what they learn and build back into their company's own software platform so the next customer benefits too.

Yamashita sets this against Japan's traditional system-engineer model. Classic SI work assumes the customer can already describe its problem in an RFP and bills by headcount (the "person-month" model). FDEs start where the customer can't yet name the problem, own everything from requirements to production, and are judged on what changed for the customer. Their employers earn from software usage fees rather than billable hours. One former Palantir FDE he interviewed said about 80% of the job is finding and surfacing tacit knowledge inside the client.

The headline point is that FDEs are often imagined as all-round "supermen", but they aren't. Yamashita lists four conditions for success: individual qualities (he names seven, including perception, communication and grit), teamwork, organisation, and widening influence step by step through small early wins. Strong FDE organisations are teams whose members cover for each other.

Why it matters: Japanese IT firms are moving quickly into FDE-style services. For engineers weighing the switch, this is a useful reminder that the job is less about raw coding and more about digging into messy customer realities. Yamashita argues that because AI now lets anyone build generic software, building something that truly fits a customer takes stubborn persistence.