If you lead a data team, you've probably heard this before: "We're drowning in requests. We need more people."

And it makes sense on the surface. When demand outpaces supply, you add supply. But here's the thing: Hiring more people doesn't fix the bottleneck. It just makes it bigger.

The real problem is how your BI tool is built.

The request treadmill

Here's a pattern most data teams know well. A stakeholder needs a report. They Slack the data team. Someone picks it up, builds the dashboard, shares the link. Stakeholder says: "Great, but can you just add one more filter?" Repeat until everyone's exhausted and nothing strategic gets shipped.

This is the request treadmill. And the cruel irony is that the better your team gets at responding, the more requests roll in. You've accidentally built a concierge service.

So you try the obvious fixes:

• Hire more people. This just means the treadmill gets bigger.

• Implement self-serve. Noble goal, but the moment a business user hits a metric they don't recognise, they're back in your Slack.

• Document everything. Also noble, but documentation goes stale the second the underlying data changes.

None of these fixes work, because none of them address the actual problem.

Your tooling is the real bottleneck

Most BI tools were designed for a world where a human does every step of the work. A dashboard is a UI artefact: a collection of clicks, not a file. A chart lives in a database of menus, not in code. Changing a metric across 30 dashboards means opening each one, clicking through, and saving it. Thirty times. There’s no other way, because the tools were never designed for anything else.

Enter the AI/BI wave, and it seems promising. AI chatbots that write SQL and let business users ask questions in natural language without involving the data team. It’s helpful to the extent that business users can log in, ask a question, and get an answer themselves. But the moment they hit a metric they don't recognise, they're back in Slack, back in your queue, and you’re back on the treadmill.

The problem is that AI bolted onto a UI-first BI tool is still a UI-first BI tool. You've got a faster treadmill. You haven't stepped off it.

The fix: ship analytics the way you ship code

The shift that actually moves the needle is treating your BI layer the same way you treat the rest of your data stack: as code.

When dashboards are code, everything changes. Every chart, every metric, every layout lives in files. Versioned in git, reviewable in PRs, deployable through CI/CD. Your semantic layer is already defined and your BI layer extends it rather than duplicating it in a parallel system someone has to maintain by hand.

But here's where it gets genuinely exciting. When your BI layer is code, AI agents don't just answer questions. They can do the work.

Say a stakeholder comes to you and asks: "This dashboard is all in USD. Can we get it in GBP?" On a traditional BI tool, that's a day of someone's time. Click into each chart. Update. Save it. Move on to the next one.

With Lightdash, a data team member opens an AI tool (like Claude Code) and says: "Update this dashboard and change everything to GBP." While the agent handles it, the team is free to do the strategic work they were hired for.

The shift to dashboards-as-code and agent-driven workflows has been huge for us. We can update, version, and replicate analytics the same way we ship software, without spending hours clicking around the UI.

Kevin Otte, Data Scientist @ fal

This is what Agentic BI actually means. Not a chatbot that answers questions. An agent that does the build work your team currently does by hand. Bulk-updating 50 charts, replicating a dashboard for a new client, or migrating content from a legacy tool. All of it handled, while your team thinks about what matters next.

The data team structures our data and keeps our pipelines running, that's their expertise. As a business user, my job is gathering insights and turning them into ideas that move the business forward. Now I can do that without waiting. That's what Agentic BI unlocks.

Hal Sarjant, Chief Commercial Officer @ Gen H
Read the case study →

Hiring more people? Make sure you've got the right tool first

Growing a team can help. But only if the team has a tool that lets them work at a different level. If they're still on a traditional BI tool, more people just means more hands on the treadmill - more time spent on manual updates, more dashboards to maintain, and still nobody doing the strategic work.

The question worth asking isn't "how do we go faster?" It's "why are we doing this at all?"

If your semantic layer is already in code, you've done the hard work. You just need a BI layer built to take advantage of it.

Lightdash is the only Agentic BI platform built for modern data teams. Ready to step off the treadmill? Book a demo and we’ll show you how.