AI-Assisted Search

Reed
Reed
  • Updated

You can use AI-assisted search to pick search filters for you after entering a query in plain English.

Benchling’s search system is powerful, and it supports over a hundred filters. Because of this, it’s often easier to describe a more complicated query than it is to manually select all the filters. With AI-assisted search, you can describe a question, and Benchling’s AI selects all the search filters needed to answer it.

Note: This is a beta feature. To be added to the AI-assisted search beta, email ai-ml@benchling.com.

Using AI-assisted search

To use AI in a search:

  1. Click the lens icon to open the search side panel.
  2. Click Set search using AI.
  3. Enter your query into the search box and click Update search settings.

After thinking for a bit, Benchling AI sets the search filters and other configuration based on your prompt.

ai_assisted_search.gif

Example queries

Here are some examples of queries you can try (these might not make sense depending on what data you have in Benchling):

  • Plasmids with kanamycin resistance
  • Notebook entries I wrote in the last month referencing CELL001
  • DNA fragments whose description contains EGFP

Limitations

The AI can use most of our search filters, but not all of them. It is good with

  • General filters (name, created at, etc)
  • Schema field filters
  • Notebook review
  • Molecular biology filters, like nucleotide string or annotation name
  • Most inventory filters

It does not support workflow filters at this time.

AI-powered features can also make mistakes, and aren’t always predictable. It will try and choose the best filters to answer the prompt you give it, but sometimes it won’t pick the right ones. Always double-check the filters it selected to ensure it gives you the results you want.

For more information about security and privacy for AI-powered features in general, visit our AI security and privacy policy.

Was this article helpful?

Have more questions? Submit a request