Evidence

Everything in Evidence, from queries to layout to theme, is defined in files inside a repo. That means your analytics get the same tools as your software: branching, diffs, peer review, and rollback. If a number looks wrong in production, you can trace it back to the exact change that introduced it. For teams already using Git, this removes a whole class of "who edited this dashboard" problems. No more guessing.

On Studio, the Evidence Agent is an AI assistant for your project. It reads your existing dashboards and reports as context, then answers questions in natural language and writes new analyses in Evidence Markdown. It respects your access rules, so readers only ever see data they're allowed to. The agent is reachable in Slack or through any MCP client, including Claude Desktop and ChatGPT, so questions get answered where your team already talks. No one waits on a queue. That's self-serve analytics in practice. Why does that matter? Because answers arrive in seconds, not days.

Evidence has a library of components for charts, data tables, filters, dropdowns, maps, and toggles. You can build publication-quality visualizations, then abstract a custom visual into a reusable component for the whole team. Rich tables support sparklines, subtotals, and images, and maps handle heatmaps and custom GeoJSON. The payoff is that a one-off chart can become a shared building block instead of being redrawn every time. Less busywork.

Push to deploy is the default. Merge your changes and they ship to users; branch previews let you see a change live before it reaches production; instant rollback reverts a bad merge on the spot. Continuous integration runs automated checks on every pull request, so broken queries get caught before they publish. That's a real difference. Traditional BI platforms let a broken chart go live unnoticed. Not ideal.

Access rules live in a single version-controlled file, and Studio adds page-level access control down to the row. You decide who can open each page, which matters when a report mixes sensitive and general data. Enterprise adds row-level rules, SSO, SCIM directory sync, and a SOC 2 report, So the same platform can serve an internal team and an externally facing dashboard without a second tool.

Any page can be delivered to inboxes daily, weekly, or monthly, and every report exports to PDF or multi-tabbed XLSX workbooks. For teams that still live in email and spreadsheets, that closes the loop between a code-built dashboard and the format a reader actually wants. It's a small feature, but often the one that makes adoption stick. Readers never open the repo. They just want the file.