Skip to main content

Search full source records, clearer memory evidence + 2 more

Find exact facts even when they were not extracted into a memory​

Chat can now search the full text of source records for specific words and phrases. This helps with precise factual questions where an important detail exists in the original transcript or document but was not included in an extracted memory.

Details: Keyword search covers speaker blocks and Markdown sections. Results include the surrounding record context and link directly to the matching paragraph, so you can verify the answer against its source. When the requested term does not occur in the records, NEXT says so explicitly instead of returning a loosely related result.

How to use:

  1. Select mode "Record extracts" in chat
  2. Ask a precise question in Chat and include the distinctive word, name, or phrase you want to find.
  3. Open a source link in the answer to inspect the matching paragraph in its original record.

Learn more about working with chat modes.

Understand each memory and trace it back to its source​

The memory dialog now makes the insight itself easier to read while keeping its supporting evidence clearly separated and traceable.

Details: A memory’s title and description are now the primary content. Supporting evidence appears in a dedicated section where every item identifies its evidence type and source. You can open an evidence item in context and follow it back to the underlying record or imported data.

How to use: Open a memory and review its Evidence section. Select an evidence item when you want to inspect the original context behind the insight.

Choose the right workflow variable with confidence​

Variable suggestions now explain what each value contains and when to use it, removing guesswork when connecting workflow steps.

Details: Suggestions include a plain-language description beneath every variable name and are available for both new and existing workflow steps. Variables that have already been inserted also reveal their description on hover.

How to use: Start inserting a variable in a supported workflow field, review the descriptions in the suggestion menu, and select the output you need. Hover over an existing placeholder whenever you need a reminder of what it provides.

Learn more about workflow variables.

Format workflow values without adding extra steps​

Workflow placeholders now support Liquid filters, letting you format and transform values directly where you use them.

Details: Apply filters in workflow fields that accept placeholders, including emails, AI prompts, and names. You can format dates, provide fallback values, change text case, trim or shorten text, replace values, and chain several transformations together. Variable suggestions and validation continue to work inside Liquid templates.

How to use:

  1. Insert a workflow variable into a supported field.
  2. Add a pipe followed by a Liquid filter—for example, {{ owner | default: 'Unknown' }} or {{ today | date: '%d %B %Y' }}.
  3. Run the step and check the rendered value in the workflow execution.

Variable names remain case-sensitive, and a step can only use output from its trigger or completed upstream steps. See the full list of examples in Use Variables in Workflows.