GoKawiilCockroach Labs says AI agents added Db2 support to MOLT in two days
Cockroach Labs reports that an autonomous system of coding agents built Db2 database migration support for its MOLT tool between a Wednesday and Friday in April, without any human writing code. The work included a new schema converter, data-fetching and verification paths, and over ten thousand lines of test fixtures, costing $4,172 in token usage. The company says this compares to nine months and roughly $160,000 in engineering time spent adding similar Oracle support in 2024.
GoKawiil's interpretation of the reporting above, not reported fact.
The comparison, if representative, suggests AI agent systems could sharply cut the cost and time of specialized engineering work like database driver support, which has traditionally required deep domain expertise. Cockroach Labs frames the result as evidence that a structured, self-checking agent workflow — which it likens to a 'coding hospital' — can maintain quality comparable to human-built features, measured by test coverage. However, this is a single internal case study from the vendor itself, so broader claims about cost and speed gains would need independent verification across other projects.
- MOLT gained IBM Db2 migration support in under two days, built entirely by AI coding agents.
- The $4,172 token cost compares to about $160,000 spent on similar Oracle support work in 2024, per Cockroach Labs.
- A planning agent broke the task into 32 sub-issues, with 27 merged pull requests and 55 review revisions before completion.
Source: cockroachlabs.com, 2026-10-09
Published there as: “Five months treating bugs like patients and coding agents like a medical team”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.