Media Publishing Operations

Reels Power-Law Performance Auditor

Audits short-form video performance across a roster of creator accounts using a power-law model. Each Monday it loads the last 28 days of reels from Airtable, computes each video's view multiple against the account's trailing median, classifies accounts as Strong, Normal, or Weak, and isolates viral outliers above a configurable multiplier. For each outlier it extracts frames, transcribes the on-screen captions, and groups the opening lines into hook frameworks (curiosity, contrarian, listicle, story, demo). The agent ships one HTML email per run with per-account tier tables, outlier cards including thumbnail, caption, and detected hook, and a roll-up of hook frameworks that worked this period.

1 runs · 1 clones

Try this agent See agent signals

Agent snapshot

Built for media publishing work, not a generic template.

runs observed

community clones

integrations used

labels matched

Media Publishing Operations

Sample input

Build an agent to reels Power-Law Performance Auditor. Use Google Sheets and Gmail. Focus on operations. Return a ready-to-review result after each run.

Sample output

Every Monday, pulls reel metrics across all managed creator accounts from Airtable, scores videos against each account's own median, tiers accounts Strong/Normal/Weak, flags viral outliers, transcribes their burned-in captions, and emails an HTML report with hook frameworks.

What this agent automates

This run includes 8 concrete steps from the published agent data.

Signal 1 · 1

Read every row from the configured Airtable base/view containing reels, accounts, view counts, post dates, and video URLs published in the last lookback window

Signal 2 · 2

Group reels by account, compute the trailing median views per account, then compute each reel's view multiple (views / account_median)

Signal 3 · 3

Tier each account by its top-quartile multiple: Strong above the strong threshold, Weak below the weak threshold, otherwise Normal

Signal 4 · 4

Select reels whose multiple is above the viral threshold, dedupe against reels already analyzed in the agent database, and cap at the per-run outlier limit

Signal 5 · 5

For each new outlier, scrape the video URL, sample evenly-spaced frames, run OCR on the burned-in captions, and stitch them into the full subtitle script

Signal 6 · 6

Use Twin AI to label the opening line of each outlier with a hook framework from the configured list

Integrations used

Reels Power-Law Performance Auditor connects Google Sheets and Gmail.

Google Sheets

Connected in this agent

Gmail

Connected in this agent

Clone Reels Power-Law Performance Auditor

Start from the exact agent behind this page, then adapt the prompt, apps, and review steps to your media publishing workflow.

Try this agent

Plans start at €20/mo