AI filters analyze content based on a natural language description you provide, rather than matching on exact keywords. Instead of monitoring for a specific word or phrase, you describe what you are looking for, and the system evaluates new content against your criteria.
Note: AI filters incur LLM token usage costs, billed separately in addition to your subscription tier cost. Gold and Platinum plans include the first $10 of token usage each month; usage beyond that is billed. Usage is typically invoiced during the first week of each month. You can review your usage on the AI Filtering LLM Usage & Billing page.
Content reaches an AI filter through two paths:
ai="filter name" flag to any of your keyword alerts. When that keyword matches, the content is forwarded to the named AI filter for deeper analysis. This lets you use fast keyword matching as a first filter before AI evaluation.Both paths can be used together on the same AI filter.
Create and manage AI filters from the AI Filtering Dashboard. Each AI filter has the following fields:
| AI Filter Name | A unique name for the AI filter (1-32 characters). This name is also used when linking keyword alerts via the ai= flag. |
| Description | A natural language description of what you are looking for (50-1,000 characters). Be as specific as possible. The AI evaluates every piece of content against this description to decide if it is a match. |
| Content Sources | Select which sites and content types (posts, comments) to monitor directly. |
| Subreddits | If Reddit is selected, you can specify which subreddits to monitor. Enter subreddit names separated by spaces, without the /r/ prefix. If left blank and Reddit is selected, no Reddit content will be monitored directly (though content can still arrive via the ai= flag on keyword alerts). |
The description is the most important part of an AI filter. The AI uses it as its sole criteria for deciding whether content is a match. A vague description will produce noisy results; a specific one will produce precise matches.
Tips:
The following options are available when creating or editing an AI filter:
Instant | Bypass your account's email delay and send the alert immediately when a match is found. |
Full Text | Include the full text of a matching post or comment in the alert, instead of a short excerpt. |
Group | Assign the alert to a named group. Alerts in the same group are bundled into a separate email. Group names also let you filter alerts on the dashboard and in RSS/JSON feeds. |
ai= Flag
You can route content from your keyword alerts into an AI filter by adding the ai= flag. When a keyword alert matches, the matched content is forwarded to the named AI filter for AI analysis. The content is only delivered to you if the AI confirms it meets the filter's criteria.
(semantic= is a still-supported alias for ai=; the two are interchangeable.)
Example
ai= flag).In this example, any post or comment containing "project management" is forwarded to the AI filter named "PM Tool Leads". The AI then evaluates whether the content matches the description you set on that AI filter. Only confirmed matches generate a notification.
If the AI filter name contains spaces, it must be enclosed in double quotes (e.g., ai="pm tool leads"). For single-word names, quotes are not necessary (e.g., ai=leads).
You can include more than one ai= flag on a single keyword alert to route the same match to multiple AI filters. For example, ai=leads ai=competitors forwards every match to both the "leads" and "competitors" AI filters, and each evaluates the content independently. Each destination runs its own LLM evaluation, and those LLM tokens are billed directly, so a keyword with two ai= flags uses roughly twice the tokens of one with a single flag.
| Gold | Platinum | Ultra (legacy) | |
|---|---|---|---|
| AI Filters | 2 | 5 | 2 |
| Subreddits Per Filter | 5 | 5 | 5 |
AI filters incur LLM token usage costs, which are billed separately in addition to your subscription tier cost. Gold and Platinum plans include the first $10 of token usage each month. You can review your usage on the AI Filtering LLM Usage & Billing page.