Saltar al contenido principal
Available on the Corporate plan. This feature is part of woku’s enterprise capabilities. Talk to our sales team.
Data Studio builds interactive data apps from a conversation. You describe what you need to see and an AI agent assembles an app with filters, indicators, charts and tables that query your company data live: whenever someone opens it, they see the current numbers. The app is not a fixed report. You can filter it, search, click an element to see its detail, ask for adjustments in the chat or edit it directly on the canvas, go back to any earlier version and share it with a key protected link.

How to use it

  1. Tell it what you want to see. For example: “a ranking of my wokus by rating, with at least 10 reviews and a folder filter”.
  2. The agent grounds the request in your data. It checks which folders, tools and periods have data, and when something is ambiguous it asks up to three short questions with options.
  3. Review the sketch. Before building, the agent draws a low fidelity sketch with the structure of the app. You can correct it in the chat.
  4. Build. Once you approve the sketch, the app is built with live data and the agent shows you the list of requirements: what it met, what it did not, and why.
  5. Adjust. Each change you ask for (“add a channel filter”, “make this chart an area chart”) modifies only what you asked; everything else stays identical.

What an app can have

Tables and lists show 10 rows per page. Clicking an element of a chart can filter the rest of the app or open its detail. Every component explains where its number comes from: the query, the filters applied and the data. When no catalog component fits what you ask for, the agent can propose a custom component, one at a time and never the whole app. It is added only if you approve it, and it runs isolated, with no access to the network or the rest of the page.

Your data and woku data

The app queries woku’s primary data: woku reviews, NPS, CSAT, CES and form responses, comments with their type and sentiment, clients, folders and tools. It also uses the comment topics found by review intelligence, so questions like “what do customers value most?” are answered by counting mentions per topic, not by the agent’s opinion. Queries are always read only and scoped to your company. You can also upload your own data in the conversation:
  • Data files in CSV, XLSX, JSON or XML, up to 25 MB and 100,000 rows. Each file becomes a table the app queries just like woku data, and it can be joined with it through a shared column, for example goals per folder against actual reviews.
  • Images in PNG, JPG, WEBP or still GIF, up to 10 MB, as a reference (“make it look like this sketch”) or to show them in the app.
  • Documents up to 50 MB: PDF, Word (DOC, DOCX, ODT, RTF, Pages), presentations (PPT, PPTX, Keynote), spreadsheets other than CSV or XLSX (XLS, TSV) and text files (TXT, Markdown, HTML, JSON, XML and others). The agent reads the whole document; for a PDF it also sees its pages. It can use it as context or a reference, or take from it what the dashboard needs.
When the app needs data from a file that is not a clean table (goals inside a PDF, a JSON with nested lists, an Excel file with several sheets or with the headers further down), the agent analyzes the file with code: a Python program reads it and writes the tables the app queries. Values come from the file through code, never typed by the AI, and the agent tells you which tables it prepared. You can attach several files in one message and combine them, for example a sales file, the logo and a document with what the committee asks for. The agent keeps seeing the files in the next messages, even if it asks you a question before building. If you upload a file with the same name again, the table gets a new version and the apps that use it show the new rows.

Templates

For the most common requests the agent starts from a template and adapts it to your request: executive summary, comment explorer, ranking, evolution and tool comparison. A template is only a starting point; the final app answers what you asked for.

Versions and publishing

  • Every change, by the agent or by you, creates a new version. No version is ever deleted.
  • You can undo and redo, compare a version with the previous one (new and changed components are highlighted) and go back to any version.
  • You can duplicate an app to explore a variant without touching the original.
  • Publishing puts a version on the shared link. The link always asks for a key, and the filters you pick stay in the URL so you can share an exact view. Publishing another version changes what the link shows without changing its address.
  • Deleting a project deletes its app and the files you uploaded. If the app was published, its shared link stops working.

Live data

Apps query the data when they open and refresh every minute while the tab is visible. Queries are cached for 60 seconds and shared by everyone looking at the same app, so many visits do not load the database.

Personal data

In the admin panel and on the key protected shared link, tables can show personal data of your clients, such as name or email. The agent never sees it: it works with fields and counts. Queries through the API and the MCP server never return personal data either.

Earlier projects

Projects created before this version of Data Studio keep their report and their published link. New projects are always created with interactive apps.

API and agents

Data Studio apps can also be read and edited without the admin panel:
  • /v1 API: list apps, read their spec and versions, run their queries, edit them with operations, build a sketch, go back to a version, copy a component from an earlier version, duplicate, publish and stop sharing, and manage the data tables. See the API reference, data-apps section.
  • MCP server: the same actions as tools (get_studio_app, run_studio_app_queries, edit_studio_app, build_studio_app, restore_studio_app_version, publish_studio_app_version, unpublish_studio_app, copy_studio_app_component, duplicate_studio_app, list_studio_app_tables and upload_studio_app_table). See Tools.
Every change through the API or the MCP creates a new version, just like in the panel.

Credits

Each message to the agent uses 1 Data Studio credit. Viewing, filtering or sharing an app uses no credits.
Última modificación el 29 de octubre de 2026