Skip to main content
Solved

Box AI for Metadata - Near and Long-Term Considerations


  • New Participant
  • 1 reply

Hi,

I think I understand the basics of metadata, as a way to connect documents that would otherwise not connect be connected (say in a folder structure or via search). And the prospect of having AI extract metadata is huge, given that having humans doing this is laborious and error-prone, and thus a difficult and predictably under-used part of Box use until now.

What do you (for example, metadata and AI product managers and power users) see as the role of metadata in an era where AI itself can now explore a storehouse of unstructured data via prompts, set up workflows and more.  Realizing there are still likely unknowns, is the innovation of AI metadata extraction something that has a specific purpose now and for the foreseeable future, or might continuing AI developments preclude its value at some point (maybe 5-10 years out, or maybe sooner?).   

I’m not meaning to take the glitter off the gold of this fantastic step forward in metadata assignment…it is likely more about my lack of knowledge of how metadata will have value even as the power of AI continues to advance. I have a few interesting use cases with Box, potentially utilizing metadata, and it’ll help me to get a better understanding of how the two interact as I work through use cases.   

Thank you!

Best answer by Meena Ganesh Box

Great question ​@TomC ! We just published our POV around why metadata matters in this AI-first era, check it out!

The TL;DR is: AI-powered metadata extraction is a transformative solution that helps organizations reveal content relationships, enhance context understanding, and streamline workflows, leading to:

➡️ Reduced compliance risks
➡️ Improved operational efficiency
➡️ Smarter decision-making

~Meena.

View original
Did this topic help you find an answer to your question?

3 replies

Meena Ganesh Box

Great question ​@TomC ! We just published our POV around why metadata matters in this AI-first era, check it out!

The TL;DR is: AI-powered metadata extraction is a transformative solution that helps organizations reveal content relationships, enhance context understanding, and streamline workflows, leading to:

➡️ Reduced compliance risks
➡️ Improved operational efficiency
➡️ Smarter decision-making

~Meena.


  • Author
  • New Participant
  • 1 reply
  • March 7, 2025

Hi ​@Meena Ganesh Box Your blogpost on this topic connected most of the dots for me. Thank you! And thanks to ​@Thomas Deely Box for the connection to you on this. 

Also commented to your LinkedIn post on this topic:
https://www.linkedin.com/feed/update/urn:li:ugcPost:7303153028801433601


thomasdeely Box

@TomC great question... ​@Meena Ganesh Box provides a comprehensive outline of where metadata is going in her blog. Further thoughts below and feel free to continue the conversation here. 

First, the difference between prompts and metadata.

  • Prompts are more about getting better “output” from AI, for example, asking AI to create a blog or a press release, for example. 
  • Metadata is often about providing better “input” to AI. The more structured input will result in better AI models and outputs. Metadata can also help with structuring workflows for example with Box Apps
  • To take an example, let's say, with supplier contracts, metadata could organize the contracts, based on supplier type, payment schedules, etc. Metadata could also help trigger workflows on the contracts, based on payment due dates for examples. Whereas prompts could be used for “output” to summarize the contracts or analyze for specific insights.

 

I see the relationship between Metadata and AI as symbiotic, metadata makes AI smarter, and AI can also create its own metadata, in a complementary relationship.

 

Even in the future, which is difficult to predict, it is always likely AI will perform better with more structured inputs and metadata will therefore continue to play a role in improving AI, even if more and more that metadata is AI generated, and less from humans, for example, as outlined in the blog. 

 

In the context of Box, Metadata is just one way of structuring data, in our recent roundtable with Baylor University  ​@Brent_Harris  and ​@Micah  mentioned they are exploring designing folder structures as a way to improve the AI, and folder structure can be seen as a form of metadata. Would love to hear more from the team at Baylor on this...

 

On March 18th, our CS team ​@NickAtBox are running a 30 minute enablement session on Metadata, signup here 

 

This space is evolving fast, has lots of nuances and lets continue the discovery here and we can loop in more of the team as necessary.


Reply


Cookie policy

We use cookies to enhance and personalize your experience. If you accept you agree to our full cookie policy. Learn more about our cookies.

 
Cookie settings