Outcomes first, tools second
Before any analytics team picks up Claude Code or Cursor, the team needs to be clear on what it actually wants to produce. Higher quality pipelines? Faster iteration on models? Better deployment hygiene? AI amplifies whatever direction you point it in. If the direction is fuzzy, you get a lot of fuzzy output, very quickly.
The takeaway: align on outcomes before you align on tools.
Low code is done
For years, the data tooling market sold productivity through constrained UIs and drag and drop builders. That trade off (give up flexibility, get speed) made sense when the alternative was writing everything by hand.
It does not make sense anymore.
When you can use Claude Code inside Meltano to build a pipeline, add a connection, and then reason about what the AI produced, with full change control and a real deployment path, the constrained UI loses on every axis. You get speed and flexibility. You get to inspect the code, version it, test it, and ship it like any other piece of software.
That is a different category of productivity.
Where AI actually belongs in your data stack
One of the strongest arguments in the video: AI should be stringing robust things together, not building from raw scratch.
You would not ask an AI to build your analytics layer from the ground up. The same logic applies to your ingestion layer. AI shines when it is connecting pre-built, well-tested components, like Meltano’s library of connectors, into pipelines that solve a specific business problem. The robust-to-robust pattern is where agentic tools genuinely accelerate teams. Building connectors from scratch with an LLM is where teams burn time and trust.
What your data team looks like in 12 months
Aaron is direct on this: your team is changing, whether you plan for it or not. AI is augmenting and in some cases replacing parts of the workflow. But the shape of a small, high-performing data team probably looks similar a year from now. Same headcount. Sharper output. Each contributor able to own a project end to end, at a quality bar that used to need three people.
If you are hoping AI will let you cut a role, you are probably going to be disappointed. If you are hoping AI will let your existing team ship better work, faster, you are pointed in the right direction.
Watch the full conversation
The video runs under four minutes and is worth the time if you lead a data team or are thinking about how AI fits into your stack.
Want to see what AI plus a proper data engineering platform actually looks like? Try Meltano with Claude Code and build your first pipeline in minutes.
Visit meltano.com to learn more