We spent 20 years on the other side of this problem
The three of us built Just Analytics into an award-winning data consultancy, and took it through an acquisition that formed the core of a global cloud data services business. Between us that is more than 60 years of turning complex data estates into governed, production-grade platforms across the major clouds.
That work taught us the same lesson on every engagement: the tools were rarely the bottleneck. What held teams back was the operating model — human-heavy delivery, discipline that depended on the most experienced person in the room being available, and knowledge that lived in people rather than in systems.
Then agents arrived and raised both bars at once. An agent consuming bad data makes wrong decisions at machine speed. An agent waiting on a pipeline blocks every initiative behind it. A data team that took two to four weeks per pipeline became the ceiling on how fast the whole company could adopt agents.
A general coding agent is not enough for that job. It has no isolated place to be wrong on real data, no guardrails that understand a data platform, no context beyond the repository, and no answer for the fact that every team runs a different platform. So we built those four things, and put a coding agent specialized for data engineering inside them.
