Learn how Vistazo works

Vistazo is a dashboard tool. You describe the chart you want and the built-in assistant builds it over your own data sources. It goes in three steps.

1. Connect a source. Five public sources are hosted for you and need no setup, and any server that supports MCP, an open standard for connecting software to AI agents, connects from Settings. See Connections.
2. Ask for a chart in chat. The assistant calls your sources, shapes the data and shows a draft in the conversation. Nothing is saved until you add it. See Chat.
3. Add it to a dashboard. The draft becomes a widget that refreshes on the dashboard's schedule, stays wired to the source it came from, and moves, resizes or restyles when you ask. See Dashboards.

The assistant works from each source's own tool descriptions, so a new source needs no modelling before the first chart.

Three kinds of thing.

• A dashboard is a page of widgets, each a chart or table that refreshes on a schedule and opens to anyone you give the address to. A parameter is a filter every widget on the dashboard can read, such as a date range. See Dashboards, Widgets and Parameters.
• A notebook runs Python over the same sources, for the questions a chart cannot answer. See Notebooks.
• A skill is a procedure written down once that the assistant and your agent follow every time. See Skills.

You build them from two inputs. A source is anything the assistant can read, whether one of the five hosted public sources, your own MCP server or a spreadsheet dropped into a chat. A chat is where you ask for charts and see them previewed. See Sources and Chat.

No. Vistazo serves the same tools over MCP, so everything the built-in assistant does at app.vistazo.ai, your own agent can do from wherever you already work.

• Make your first dashboard walks the website from sign-in to a shared dashboard.
• Connect your own agent adds Vistazo to Claude, Claude Code, Cursor, Codex or any MCP client. Vistazo as an MCP server covers what that agent can reach.