Introducing the Wren AI Gallery

Every dashboard starts the same way: a blank page and a vague sense of what "good" looks like. The Wren AI Gallery removes that blank page. It is a public library of finished, working analytics — and every one of them shows its work.

Start from a report that already works

The Gallery is a growing set of real dashboards and reports across banking, e-commerce, SaaS, healthcare, oil & gas, and manufacturing. Each one was built by the Wren AI agent from a single prompt — a risk desk view of banking transactions, a SaaS revenue and churn report, a hospital executive dashboard, a production monitoring view for a factory floor.

These are not mockups with hand-picked numbers. Each entry is a working GenBI app the agent actually produced, running on a real dataset, with the reasoning that produced it preserved end to end.

Open any example and you'll see the prompt that produced it and every step the agent took to get there: which models it searched, how it resolved the business terms, the SQL it planned, and how it assembled the result into an app. Like what you see? Point the same template at your own warehouse and the agent rebuilds it on your data, running the SQL under your existing access policy.

Here's what reusing a template looks like in practice:

What you get

• The whole process, not just the output — read the exact prompt and every step the agent took, from model search to final chart. Nothing is hidden behind "AI magic."
• Reuse any template — retarget it to your own warehouse in a click. The agent replays the analysis against your schema and your definitions, not a frozen copy of ours.
• Your guardrails stay put — the agent runs SQL under your existing access policy. Row- and column-level security applies at query time to the person asking, and every rebuilt report carries a full trace.

Why we built it

Two reasons.

First, the fastest way to trust an AI analyst is to watch it work. A demo video shows you the happy path; the Gallery shows you the actual reasoning — real prompts, real steps, real SQL — on datasets complicated enough to matter. If the agent's process doesn't hold up to inspection, you'll see it. We think it does, and we'd rather you judge that yourself than take our word for it.

Second, most analytics needs aren't unique. A churn report, an inventory view, a patient-demographics breakdown — thousands of teams need the same shape of answer over their own data. Starting from a template that already works turns "build me a dashboard" from an open-ended project into a one-prompt adaptation. The agent does the retargeting; the context layer makes sure the result means what your organization means.

The Gallery launches with examples across six industries, and it grows with every release. If there's a report you'd like to see the agent build, tell us — the best templates come from real questions.

Browse the Gallery →