// USE CASES · §06

What you build with it.

Six concrete workflows the entit.ai API is built for — from brand monitoring to research at panel scale. Each one maps to the default VideoAnalysis or one of the 24 workflow formats so you skip the client-side parsing.

// HOW TO READ THE PANELS

Each panel lists a recommended format (or the default analysis) for the job. Same POST /v1/jobs endpoint, same price — just a different shape in result. Full catalog: workflow formats.

01│BRAND

Brand monitoring across TikTok, Reels & Shorts

Find every video where your brand or product shows up — not just the ones that tag you. entit.ai surfaces mentions in spoken audio, on-screen captions, and visual context (a logo on a hoodie, a packshot on a counter), all in one structured response.

// RECOMMENDED FORMAT

(default · VideoAnalysis)Sweep feeds for mentions across transcription, on-screen text, and visual context.
format: content-screeningDecision-grade brand-safety verdict before reposting or buying UGC.

// PROMPT / QUERY

Search transcription, onScreenText, and visualDescription for 'Brand X' across the last 7 days. Group results by sentiment and creator handle.

EXCERPT · BRAND

SELECT video_id, author_handle, processed_at
FROM videos
WHERE transcription ILIKE '%brand x%'
   OR on_screen_text ILIKE '%brand x%'
   OR visual_description ILIKE '%brand x%'
ORDER BY processed_at DESC

02│TRENDS

Trend, sound & format detection before the peak

Cluster visualDescription and transcription embeddings across thousands of new posts a day. Spot emerging topics, sounds, and formats in their early-mover window — not after the trend has saturated your feed.

// RECOMMENDED FORMAT

(default · VideoAnalysis)Best for clustering at volume — full transcript and visual narrative for embeddings.
format: sound-trendUse on candidate clips to score viral potential and audio context.

// PROMPT / QUERY

Cluster this week's analyses by embedding(visualDescription). Surface clusters growing >40% week-over-week with sample videoIds.

EXCERPT · TRENDS

WITH this_week AS (
  SELECT video_id, embedding
  FROM videos
  WHERE processed_at > NOW() - INTERVAL '7 days'
)
SELECT cluster_id, COUNT(*) AS volume
FROM cluster(this_week)
GROUP BY cluster_id
ORDER BY volume DESC

03│UGC

Customer UGC analysis & review mining

Customers don't write reviews any more — they post videos. Pipe UGC about your product through entit.ai and get back a structured transcript of what they love, what broke, and which features they actually use. Beat your own NPS surveys with real talk from creators.

// RECOMMENDED FORMAT

format: review-summaryTyped pros/cons + recommendation per video — no client-side claim extraction.
format: content-screeningLayer in when you also need a brand-safety check before featuring the UGC.

// PROMPT / QUERY

Submit each UGC clip with format: review-summary to get pros/cons and a recommendation, ready to merge into your product feedback pipeline.

EXCERPT · UGC

{
  "pros": ["clear sound for the price"],
  "cons": ["battery dies after one day"],
  "recommendation": "mixed",
  "sentiment": "neutral",
  "verdict": "Solid budget pick if you charge nightly."
}

04│CREATOR

Influencer vetting & competitor intelligence

Before you sign a partnership, find out what a creator actually talks about across their last hundred posts — not just their pinned reel. Same trick for competitor accounts: track topic shifts, on-screen claims, and production-style changes week over week.

// RECOMMENDED FORMAT

format: creator-briefHooks, key moments, and the vetting subobject in one paid call per video.
format: content-screeningPair with creator-brief when the partnership is brand-sensitive.

// PROMPT / QUERY

Submit each video with format: creator-brief. The vetting block surfaces themes, audience fit, brand-risk flags, and a partnership recommendation per clip.

EXCERPT · CREATOR

{
  "hook": "I grew 50k followers in 30 days...",
  "angle": "Organic growth through short-form hooks",
  "vetting": {
    "contentThemes": ["creator growth", "social media"],
    "tone": "energetic, instructional",
    "brandRiskFlags": [],
    "partnershipRecommendation": "approve"
  }
}

05│REPURPOSE

Repurpose short-form video into long-form content

Turn every Reel and Short your team ships into a blog post, newsletter section, or accessible transcript. Stable transcription plus scene timestamps gives your editorial pipeline clean source material with citations back to the original clip.

// RECOMMENDED FORMAT

format: tutorial-guideHow-tos → written guide / FAQ structure ready for the CMS.
format: quote-highlightsPull quotes and timestamps for newsletters and social cards.
(default · VideoAnalysis)Use when you need the raw transcript + scenes[] for editorial work.

// PROMPT / QUERY

Submit how-tos with format: tutorial-guide for written steps, or format: quote-highlights to surface the most shareable pull quotes with timestamps.

EXCERPT · REPURPOSE

{
  "title": "How creators are using on-device AI",
  "highlights": [
    {
      "timestamp": "0:42",
      "quote": "Consistency beats virality every single time.",
      "shareability": "high"
    }
  ]
}

06│RESEARCH

Research, journalism & market intelligence at scale

Academic teams, newsrooms, and market-research firms use entit.ai to process thousands of social videos without scraping. Stable JSON means downstream sentiment, language, and behavioural analysis doesn't break the next time a platform changes its DOM.

// RECOMMENDED FORMAT

(default · VideoAnalysis)Canonical full-fidelity record — best for downstream NLP / coding frames.
format: eyewitness-timelineNewsrooms: structured timeline of factual claims for citizen-journalism intake.
format: health-claimsPublic-health and fact-check teams: flag medical claims for review.

// PROMPT / QUERY

Across 5,000 analyses, count language distribution and the top 50 lemmatised nouns in transcription. Bucket by week for trend lines.

EXCERPT · RESEARCH

SELECT
  language,
  date_trunc('week', processed_at) AS wk,
  COUNT(*) AS n
FROM videos
GROUP BY language, wk
ORDER BY wk DESC

// MORE APPLICATIONS · §06.B

Who else is building on it.

The same structured response unlocks workflows across marketing, publishing, commerce, and research. A non-exhaustive map of what teams are wiring entit.ai into today.

A│CATEGORY

Social media & marketing

Agents and analysts watching what gets said, not just what gets viewed.

→Brand monitoring across TikTok, Reels and Shorts
→Competitor analysis on what content actually says vs view counts
→Trend detection for emerging topics, sounds and formats
→Influencer vetting before partnerships and campaigns

B│CATEGORY

Content & publishing

Editorial teams turning short-form into the rest of their content stack.

→Auto-generating article summaries from video content
→Repurposing short-form video into blog posts or newsletters
→Subtitles and accessibility for large content libraries

C│CATEGORY

E-commerce & product

Product teams listening to the conversation customers are already having.

→Extracting product mentions and recommendations from creator videos
→UGC analysis — what customers say about products in their own videos
→Review mining — "what are people saying about X on TikTok"

D│CATEGORY

Research

Anyone who needs to process social video at panel scale, reliably.

→Academic and social research into trends, language and behaviour
→Journalism rapidly processing large volumes of social video
→Market research firms analysing consumer sentiment

// READY TO INTEGRATE?

API reference →Playground