AI Coding Assistants Haven’t Sped up Delivery Because Coding Was Never the Bottleneck
This matters because enterprise architecture decisions around AI, data, and platform engineering define long-term competitiveness and operational efficiency.
AI Coding Assistants Haven’t Sped up Delivery Because Coding Was Never the Bottleneck
Agoda recently published an observation arguing that while AI coding tools have measurably raised individual developer output, the resulting velocity gains at the project level have been surprisingly modest, because c...
Editorial Analysis
The Agoda finding confirms what I've observed in the field: AI coding assistants are productivity theater if your delivery bottleneck isn't raw code generation. In data engineering specifically, I've seen teams ship boilerplate faster while still waiting weeks on schema reviews, data lineage clarification, and stakeholder alignment on metric definitions. The real constraint is architectural decision-making and cross-functional coordination, not typing speed. This reshapes how we should invest in AI tooling. Rather than chasing code-per-minute metrics, data teams should focus AI on the actual blocking points: automated data quality validation, intelligent lineage inference, and self-service analytics abstractions that reduce tribal knowledge. For platform engineering decisions, this means prioritizing investments in governance layers and metadata infrastructure over code-generation features. The teams accelerating delivery aren't those with the fanciest IDE plugins—they're the ones with clear data contracts, automated testing, and decoupled dependencies. If your bottleneck is still code, you have bigger architectural problems to solve first.