Users of spreadsheets and databases are already familiar with the concept of auto fill. Affordances in tools like Excel have made this capability easily discoverable for years.

It’s no surprise then that this familiar paradigm is being extended to include GenAI capabilities.

  • Before: A user of Google Sheets might insert a date in the first cell of a column and be prompted to auto-fill dates for the remaining cells in order or based off of some formula
  • Now: A user might tell the system to capture the date of incorporation for everyone company in a database, and rely on the model to return those dates automatically into the column

This pattern is a clear example of the agentive nature of Artificial Intelligence to complete mundane tasks and save us time. The implications here are especially powerful when considering large data modeling of proprietary information, and the ability to connect people with information that has been available to them but buried.

For example, imagine a Customer Success Manager creating a spreadsheet of their customers, and using Auto Fill to capture usage data, risk assessments, and the names of their key stakeholders in one place ahead of a Quarterly Business Review.

Like other uses, this request type could be susceptible to mis-information. Trust markers like confidence indicators or fingerprints such as direct links to the scrapped sources should be considered to provide for easy human-review of the results.

Details and variations

  • Auto-fill is commonly seen in spreadsheets.
  • Allow users to create an auto-fill column, or use default smart columns to make it even easier
  • When users create a new prompt for a column, show a few sample records before filling them all
  • Be mindful of overwriting existing data when applying auto-fill to a column

Considerations

Positives

Quick feedback loops of value
Unlike Open Chat, using Auto Fill as a prompt input makes it easy to draw the connection between the user's day-to-day tasks and the value that the model can provide to make their lives easier.

Contextual wayfinding
Auto Fill generally starts with some structure in place, such as columns and headers in a database, and often references existing data. It can be easier to provide Wayfinding clues that feel somewhat personalized, compared to other request types. Think about what data a user is likely to have or be willing to share, and then consider offering Auto Fill through progressive disclosure as a means of getting the user started with the technology.

Potential risks

Proprietary exposure
It's far more likely that someone would be sharing personal or propriety data in spreadsheets, compared to open-ended prompt generation. There is a reason large companies are restricting their employees from sharing this data with public LLMs. Consider the security implications of this feature.

Use when:
A single prompt can be applied over and over across a set of inputs, such as in a database or table

Examples

Notion's database AI tools allow users to set up a prompt and apply it to all records in a database as a property
Google sheets expands upon its existing auto-fill capabilities to generate contextual data base on other information
Relay allows users to generate content for fields based on a single prompt
Coda uses pattern recognition to auto-fill empty cells based on existing data
Attio includes smart columns that auto-fill preset data without a prompt
No items found.