Skip to main content

Using an Agent to solve the Content Discovery Problem

  • December 23, 2025
  • 4 replies
  • 53 views

juan.delrio
Forum|alt.badge.img+2

I started this because keeping content fresh in the mind of a constantly growing and rotating audience is impossible.  If they don’t know it exists they won’t search for it, and any lessons shared in an article older that 2 weeks means that it has a good chance not to be seen by everyone.   

I also had an issue where all my Subject Mater Experts where always too busy to create new content for the community.  

 

Then it dawned on me.  I don’t just have internal SME’s I have an entire community of SME’s with 3 years of content!   I can take an LLM to find any community question with best answer and resurface it as an article!

Now I needed an agentic solution to make this happen.  Gainsight CC doesn’t have yet.

But we all use Zapier and they just rolled out Zapier Agents.  The planets are aligning!  

After a few minutes I settled on a solution:

I created a Zapier Agent, that pics a random post for a different 3 month time period for each day of the work week. It picks a random customer question with a best answer and recreates the content as an article with a title like Monday’s Use Case Highlight. 

Because it is not creating new content, I don’t run into the LLM issues of fabrication.  It is grounded on the customer question and the internal SME best answer response. 

I cannot currently post directly into the community, but I can create article drafts that I can then schedule to post myself.  Currently building library of article drafts for me to review and verify that everything is working.  
 


Here is a sample article it just made for me:
 

​​​​​​I am thinking of finetuning the heavy handed community crediting but so far sooooo gooood.   Starting January I am incorporating this content into the community!

@DannyPancratz 

4 replies

jillian.bejtlich
Forum|alt.badge.img+2

Oh I am digging this!

We’re a few months into an AI optimization project that has similarities, except our criteria is UGC that has high impressions through Google Search Console (therefore indicating high ingestion into AI overviews) or high page views (therefore indicating high consumption through direct traffic). We similarly use AI to write an article, but ours is focused on optimizing for AI ingestion to ensure AI answers are accurate and relevant. I like how yours is optimized for human consumption.

Thank you for sharing!


jvdc
  • Helper ⭐️⭐️⭐️
  • December 23, 2025

Very interesting!


juan.delrio
Forum|alt.badge.img+2
  • Author
  • Helper ⭐️⭐️⭐️
  • December 23, 2025

@Sudhanshu ​@vishkatti 


revathimenon
Forum|alt.badge.img+8
  • Gainsight Community Manager
  • December 24, 2025

Very cool!