Skip to main content

AI Filtering

Boolean logic is great for keywords, but some things are hard to define with just "AND" and "OR". AI Filtering allows you to use natural language instructions to evaluate every match found by your monitor.

How it Works

AI Filtering acts as a secondary layer of evaluation.

  1. Your Boolean Logic finds a potential match.
  2. The match is sent to the AI Engine.
  3. The AI evaluates the content against your Instruction.
  4. The AI either Approves or Rejects the match.

Writing Instructions

You can write instructions in plain English. The more specific you are, the better the results.

Example Instructions

  • "Only approve this if the user is asking a technical question about Python. Reject if it is a meme or a low-effort post."
  • "Approve if the sentiment is negative and the user is complaining about a bug. Reject if they are asking for a feature."
  • "Only approve if the post is specifically about the 'Redd' monitoring tool and not the color red or the social network Reddit."

Cost

AI Filtering is a premium feature that uses AI Credits.

  • Cost: 1 credit per match evaluated.
  • Pay-as-you-go: Credits are only spent when a match is actually sent to the AI for evaluation.

Benefits

  • Reduce Noise: Filter out 90% of irrelevant matches that keywords alone can't catch.
  • Sentiment Analysis: Automatically flag only the posts that matter to your brand.
  • Context Awareness: Distinguish between different meanings of the same word based on the surrounding text.
Credits Required

You must have a positive credit balance for AI Filtering to run. If you run out of credits, the monitor will continue to find matches using boolean logic, but the AI layer will be skipped.