AI assistance
AI that works on your report
Point the AI at the part of your report you want changed, ask in plain language, and review the result before it lands. Every answer is grounded in your data.
The Context Lens
The Context Lens is the single place AI lives in the editor. Open it with ⌘. or ⌘K.
Open the Lens on a selection and it targets exactly that: a whole page, a chart, one series, a single bar, a table cell, a line of text. That target becomes the context for your request, so the AI changes what you pointed at instead of guessing.
Common ways people use it:
- Ask for an edit in plain language: recolor a series, tidy a layout, rewrite a caption.
- Generate a chart from a natural-language request, bound to a dataset you have already connected.
- Explain this chart or suggest insights for a chart you are looking at.
AutoReport: draft a full report
Point AutoReport at a dataset and it drafts a complete first report (pages, charts, tables, and narrative) for you to refine.
- Connect a dataset (see Connect your data).
- Open the Lens and choose AutoReport, then set a few parameters.
- Review the generated draft and edit any part of it like any other Rahoto report.
AutoReport gives you a starting point, not a finished deliverable. You stay in control of the design, the wording, and what ships.
Reader Q&A on published reports
On a published report, readers can ask questions about what they are looking at and get an answer drawn from that report's data. Answers cite what they are based on, and the system is built to decline rather than invent: if the report does not support an answer, it says so.
Reader Q&A is part of how you publish and deliver a report, not a separate app your readers have to learn. Reader access is free, and answers are computed from the published data, so they don’t use the workspace’s AI capacity.
Grounded, private, and reversible
The AI is designed so you can trust what it produces and stay in control of your data.
- Grounded in your data. Generated charts, tables, and narrative are checked against your dataset's profile, so fields and columns that are not in your data are caught rather than passed through.
- Personal details are pseudonymised before a cloud model call. Cloud requests send your prompt with your data's structure and aggregate profiles, never raw rows, and every one passes a single policy check first. Names, postal addresses and postcodes, emails, phone, card, bank account and national ID numbers, IP addresses and dates of birth are detected on Rahoto's servers and replaced with workspace-specific placeholders. Unclassified data is treated as confidential, a check that cannot finish sends nothing, and restricted data stays on the local provider. The AI request log keeps only the redacted text. Other free text you type is sent as written.
- Previewable and undoable. AI edits show you what they will change before you accept, highlight the affected elements on the canvas, and land on the same undo stack as your own edits, so a single undo reverses them.
Good to know
A core set of AI actions, like generating a bound chart, runs without any cloud model at all. Cloud-backed features only reach an external model when your workspace is configured for one, and after the pseudonymisation above. For our full data and privacy posture, see the Trust pages.
Try it on a real report
Connect a dataset, open the Lens with ⌘K, and ask for your first change.