CLERINT Social Pulse
CLERINT SOCIAL PULSE · Keyword & narrative monitoring
Early access · limited availability

Standing watch on the words that matter.

Track the keywords, handles and phrases a case depends on across six platforms — and see how each named person is being spoken about, not just whether a post was positive.

WatchesKeywordsPhrasesHandlesHashtags
Watches
6
Platforms from one keyword set, with no official APIs and no accounts
Scores
Per entity
Sentiment toward each person named in a post, not the post overall
Reads
Video
Transcribed and OCR'd on demand, including Urdu
Alerts
Instant
Or digested — with the overflow attached as a PDF
Coverage

Six platforms, one keyword set.

No official platform APIs, and no logged-in accounts — so there is no application to be rejected and no account to be banned mid-operation.

Six platforms from one keyword set — no official platform APIs, and no logged-in accounts.

X (Twitter)Public mirrors.
BlueskyPublic API.
MastodonPublic timelines.
ThreadsPublic feeds and search.
InstagramPublic search and viewers.
TikTokPublic viewers.

Coverage of platforms without a public API depends on third-party public mirrors and is best-effort. We scope exactly what is achievable for your keywords in a briefing.

Platform names and logos are trademarks of their respective owners. Their presence here identifies the sources CLERINT can lawfully collect from, and does not imply any partnership, endorsement or affiliation.

The differentiator

Sentiment toward each person named — not toward the post.

A post can praise one figure while attacking another. Post-level polarity averages that into nothing useful. Social Pulse scores every named entity separately, so a monitor tells you how each subject is being talked about rather than how the post feels overall.

Per-entity scoring
Each person, party or organisation in a post carries its own sentiment chip.
Language identification
176 languages recognised, so a monitor is never silently English-only.
Key phrases
Surfaces what a burst of posts is actually about, without you having to read all of it.
Post-level sentiment
Scored locally on every post that passes the filter — nothing is sent out to be read.
Precision

Whole-word matching, so near-misses never enter the dataset.

A monitor is a structured query, not a bag of words. Terms combine, exclusions apply after collection, and matching is whole-word — so tracking Ata Tarar never fills the dataset with posts about Qatar.

keywords
All must appear.
phrase
An exact sequence.
any_of
Any one of these — compiled into separate searches and fanned out, so spelling variants each get a real query.
none_of
Excluded after collection, so an exclusion never narrows what was fetched.
since / until
A bounded window.
lang
Restrict to a language.
platforms[]
Which of the six to watch.
max_posts
A ceiling per platform per run.
Video

It reads what the video says, and what is written on it.

A short-form video carries its message in speech and in on-screen text, and neither is in the caption. One click transcribes and reads both, then two models independently summarise what it means.

Transcription
Multilingual speech-to-text, with Urdu, Hindi and English handled directly.
On-screen text
Read from sampled frames, catching captions burned into the video that no caption field carries.
Dual-model summary
Two models run in parallel and both answers are kept, so disagreement is visible rather than hidden.
Alerting & distribution

Built for someone reading it on a phone.

Trigger
Positive, negative, neutral or any sentiment toward a monitor's keywords, matched fuzzily so Shahbaz Sharif also catches Shehbaz Sharif.
Delivery
Instantly, or as a digest every 30 minutes, hour, 2, 6 or 12 hours, or daily.
Recipients
Several per rule, with a test send before it goes live.
Email shape
A rendered card for the lead post, the first 15 matches inline, and the remainder attached as a PDF.
Onward sharing
A WhatsApp share that pre-fills the message, and public links that render a single post standalone with its own preview image.
Deduplication
Global — the same post surfacing from two sources is stored once and alerted once.

Social Pulse is in early access. Coverage of platforms without a public API depends on third-party public mirrors and is best-effort; we scope what is achievable for your specific keywords before you commit.

Common questions

The things people ask before a briefing.

Which platforms does it monitor?

X, Bluesky, Mastodon, Threads, Instagram and TikTok, from one keyword set — with no official platform APIs and no logged-in accounts, so there is no application to be rejected and no account to be banned.

How is this different from ordinary sentiment analysis?

It scores sentiment toward each entity named in a post rather than the post as a whole. A post praising one figure while attacking another produces two different readings, not one meaningless average.

Does it read video?

Yes, on demand. A video is transcribed with Whisper — multilingual, including Urdu — and its on-screen text is read with OCR, then summarised. Both carry meaning the caption usually does not.

Is it generally available?

It is in early access. Coverage of platforms without a public API depends on third-party public mirrors and is best-effort, so we scope what is achievable for your specific keywords before you commit.

Tell us what you need to watch.

Social Pulse is in early access. Bring the keywords and handles that matter to you and we will scope exactly what is achievable across each platform before you commit to anything.