Intelligence·Apr 8, 2026·12 min read

AI Coding Assistants in 2026: Honest Productivity Numbers

Cutting through the demo theater to look at what AI tooling actually does to engineering velocity.

LW
Liam WuContributor, The Signal

After three years of breathless productivity claims, the field is finally producing serious longitudinal data. The honest summary: AI coding assistants are a real, measurable lift for greenfield work and well-documented domains; a marginal lift for mature codebases with strong internal idioms; and a negative lift for some categories of debugging where the model's confident hallucinations create new investigation work.

The productivity story is also strongly bimodal. Senior engineers use these tools as a fast typing surface for code they already understand. Less experienced engineers use them as authoritative answers, with predictable consequences. This is not a tooling problem — it is a training and review-culture problem.

The Dispatch

The Signal in your inbox

Join 42,000+ software leaders for a weekly briefing on the architectural shifts and economic trends shaping the next decade of SaaS.

No spam. One email a week. Unsubscribe at any time.