Decision models

Sort creators' DMs and comments for a monthly fee

Set up a decision model (a fast AI that only makes yes/no, choice, or score calls) to sort a creator's DMs into paid work, fans, pitches, and spam, and to tag comments as questions, complaints, or praise. Busy creators pay a monthly fee so they only see what needs them.

The hard part: connecting each creator's inbox and comments on every platform they use, and finding creators busy enough to pay.

Setup 3 daysRuns $25/moYour time A weekly check-inReadiness Readiness 66 out of 100Some detailHow we score thisNeeds Decision modeln8nAirtableStripe

Who pays

Creators with busy inboxes pay a monthly fee to have their messages sorted and paid-work offers surfaced. The poster puts the AI cost at roughly $0.63 a month for 15m input tokens, which is the poster's own figure, so nearly all of the fee is margin. No source reports what creators pay for this, so the monthly price is only a suggestion.

How to build it

  1. Connect the creator's inbox and comments

    In n8n, connect the creator's Instagram professional account through Meta's official tools and their YouTube comments through YouTube's data feed, so new messages and comments flow in automatically.

  2. Write the sorting questions

    Write one choice question for DMs (paid work, fan, pitch, or spam) and one for comments (question, complaint, or praise). Agree on the labels with the creator.

  3. Send each message to Jev

    Have n8n send each DM or comment to Jev with the matching question. Save the label it picks and the confidence score that comes with it.

  4. Set action thresholds for comments

    Decide with the creator what happens at each confidence level. The poster's example is to hide a comment at 0.7 and ban the user at 0.9.

  5. Log everything in Airtable

    Add every sorted message to an Airtable base, with a view of paid-work DMs the creator checks each day.

  6. Charge a monthly fee

    Bill each creator's monthly plan through a Stripe Payment Link.

Sources on X

  • 0xSkik · Sep 22, 2026

    0xSkik describes using TypeSafe's Jev model to route DMs into paid work, fan, pitch, and spam, and to set comment thresholds of hide at 0.7 and ban at 0.9, with every answer coming with a confidence score. The poster says Jev answers in 70–500ms, costs $0.042 per million input tokens with free output, and that at 15m input tokens a month the whole system costs roughly $0.63. These are the poster's own figures.

Ideas are curated from public posts and untested by us. Read the full note.