Conversation Context

Beyond the semantic layer and organization settings, you can add context directly within a conversation for faster, more targeted answers.

Scope your chat context

When you have a dashboard, a dataset, a specific visualization, or any file open alongside the AI chat panel, it's automatically taken in as context, so answers stay relevant to what you're seeing and working on.

You can also reference specific objects yourself (datasets, dashboards, and visualizations) to help AI narrow its focus and avoid guesswork.

• Inline reference: Type @ in your message to reference an object for that question only.
• Pin context: Click + Context to pin an object for the entire conversation.

Example: Instead of "What's our revenue this month?", try "What's our revenue this month? @ecommerce_orders_dataset". This tells AI exactly which dataset to use.

Add files and images

Coming soon

This feature is not yet available. We'll update this page when it launches.

You can attach images and plain text or code files directly into a conversation.

This is useful when you want to:

• Rebuild something that lives outside Holistics. Attach a screenshot of an external dashboard and ask the AI to recreate it.
• Build out new metrics from a spec. Attach a screenshot of the metrics your product team shared in Slack, and the AI drafts them for you to review and approve.
• Analyze against a list or a number that isn't in your warehouse. Attach this quarter's budget and compare it to actuals.

Supported file types

• Images: PNG, JPG, WebP, GIF, SVG.
• Plain text and code files: TXT, MD, HTML, CSV, JSON, XML, SQL.

Who can see your attached files

Attached files follow the same chat access permission, i.e. only you and your admins can see the attached files inside the conversation.

File storage

Attached files follow the chat's retention. When a chat is manually deleted, or automatically deleted due to the retention period, the attached files are deleted as well.

Chat memory

Holistics AI remembers your conversations from the last 10 conversations, so you don't need to re-explain context from a previous session. If you spent time yesterday exploring a revenue drop, today's follow-up questions will benefit from that prior context automatically.

The 10-conversation memory window is the current default. We plan to extend this in a future release.

Give feedback on AI answers

Use the 👍 / 👎 buttons on any AI response to signal whether the answer met your expectations. AI uses this feedback to learn your preferences across future sessions (in your personal workspace): for example, consistently downvoting verbose answers will shift it toward more concise responses over time.