Currently, while analyzing Algolia "No Result" searches, we follow a manual process:
  1. Exporting the CSV
  2. Converting it to Excel
  3. Going through the keywords & identifying documentation gaps
  4. Identifying & updating related articles
Similarly, when reviewing Eddy AI search results, we categorize them into:
  • Correct Answer
  • Wrong Answer
  • Info Available but Not Fetched
  • Documentation Gap
Feature Request
Can we build an in-product solution for this? Ideally, within the Analytics section, where we can take action directly:
Dislike Feedback Tracking – When a user dislikes a response, I want to view the citation and see where Eddy AI fetched the information from and maybe update the article if required.
AI Training & Corrections – I want the ability to train Eddy AI by providing custom answers or update specific articles where:
  • It failed to return results.
  • It provided incorrect information.
This would improve search accuracy and make the users take action on the analytics information within the product.