Changelog

You can now configure which data sources are selected by default in a new AI conversation. For those with multiple Supersimple accounts: we now also remember your source preferences on a per-account basis.

You can now ask Supersimple's AI questions through Slack using the /supersimple ask and /supersimple research slash commands.

Other improvements

• The Google Drive agent can now be directed to analyze specific files by providing the link in your prompt.
• Vastly improved multilingual support for AI, detecting your prompt's language.

Null values in charts are now displayed as gaps instead of zeros. This should make understanding missing data a bit less confusing!

When you cancel a chat request, the responses that were already streamed will now be preserved instead of being discarded.

Google Drive references now display the appropriate file type icon based on the document's mime type.

Setting up Slack integrations is now completely self-service. You can add Slack to your account and invite the bot to channels without talking to a single human. The moment you finish setup (under account settings), Supersimple will start indexing your chat history.

Note: only channels that the Supersimple bot is invited to will be indexed by default.

Categorical charts can now be flipped between horizontal and vertical layouts from under the Chart settings in the sidebar.

Other improvements

• Timeseries tooltip contents are now sorted by value (instead of key).
• The Jump to related data relation picker now lets you search for relations.

Each hovered chart now shows a quick shortcut to open up that chart's settings in the sidebar.

Google Drive search results now include footnotes to the relevant source documents.

Variable blocks in multiselect mode now include convenient All and None shortcuts to quickly select or deselect all options.

When the Supersimple Slack bot is added to a new Slack channel, it will now automatically discover and index that channel's history.

Variables whose types don't fit a certain use case are now clearly marked as disabled, and show a tooltip outlining the issue.

AI now streams in progress updates as it thinks.

Other improvements

• Improved handling of interrupted queries when you reload, navigate away, or cancel the AI's work.

Filter operators now automatically switch to match the variable type you're working with. Less manual work, hooray!

Other improvements

• We reworked our entire unstructured data retrieval system. Expect to see more relevant search results!
• When AI references the same file or link multiple times, those are now collapsed into one footnote.

AI settings now separate mode selection from data source options.

Our new Research agent can run a large number of searches across both your data warehouse & unstructured data sources. Currently only available to select accounts – contact us for access!

Other improvements

• AI responses now cite sources more naturally, using inline quotes for single sources and footnotes for multiple sources.

Users can now choose to accept or discard just the last reply or the entire conversation.

AI now remembers your last used data source(s) and defaults to that for the next message.

Other improvements

• When editing a previous message, we now default to that past message's enabled data sources.

You can now filter for fields from related hasOne models without first "bringing them into" the table via New Column.

Each relation's fields are under a relevant subheading in the Filter field dropdown.

You can now organize explorations into folders in the sidebar by prefixing their names with a parenthesised name, e.g. (My Folder Name) My Exploration.

You can now toggle both stacking and "show values" from the chart options sidebar.

Dynamic enum values are now cached for better performance, with configurable TTL (default 6 hours). See docs for more!

You can now specify data sources when asking AI a question.

Supersimple's AI can now query multiple data sources simultaneously.

Hovering explorations in the left sidebar now displays information on when they were last edited. And, even more details in the right sidebar once you open up an exploration.

The data model now supports defining operations within hasMany relationships for more complex join logic. See docs for more!

You can now click directly on chart elements to drill down into the relevant data on a row level.

The new net_work_days(start_date, end_date) function calculates the number of business days between two dates, excluding weekends.

Enum options can now be loaded dynamically from queries, instead of having to hardcode every value. See docs for more!

models:
  user:
    properties:
      country_code:
        type: Enum
        enum_options:
          load: distinct_values

New CLI commands discover bigquery enum and discover postgres enum make it easier to document enum values in your database schema.

Other improvements

• AI can now use the native date range functionality; it previously loved to use separate filters for start & end date.