Grounded in structured data, not a generic chatbot. Every summary is generated from the same NLP rules-based sentiment engine. Never a free-standing language model guessing at the story.
How it's built
From signal to sentence.
01 / AggregateThe last 24 hours, per nameSocial and news activity for a name is pulled into a single rolling window.
02 / SynthesizeThemes and catalysts identifiedThe dominant drivers behind a sentiment shift are isolated from the noise.
03 / SummarizeA plain-English summary generatedConcise and grounded in the underlying data — nothing freely invented.
04 / ScorePaired with the sentiment scoreThe summary and the number are generated from the same data, so they always agree.
Structured output
A sentence for the desk. A field for the model.
Each summary arrives with the sentiment score, drivers and source mix behind it — nothing to reverse-engineer before using it.
Plain-English summary of the last 24 hours of activityNamed drivers behind the sentiment shiftSource mix across social and news
Sample AI summary● Live
Sentiment on ACME turned sharply positive over the past 24 hours, driven by a wave of bullish StockTwits commentary following Q3 earnings and raised guidance. Discussion volume is running well above its 30-day average.
{
"ticker": "ACME",
"generated": "2026-09-29T14:31:02Z",
"sentiment": 0.82,
"summary": "Sentiment turned sharply positive over the past 24 hours, driven by bullish StockTwits commentary following Q3 earnings and raised guidance.",
"drivers": ["earnings", "guidance raise"],
"sources": ["stocktwits", "news"]
}
Illustrative sample. Field names to be confirmed against the data dictionary.
Applications
Where desks put it to work.
Morning briefingsRead what moved overnight without reading every post.
Alert triageUnderstand why a name is flagged before digging into the raw data.
Client communicationShare plain-English context with clients or stakeholders.
Research kickoffGet up to speed on a name's narrative in seconds.
Model QAConfirm why a sentiment score moved in a given direction.
FAQ
Frequently Asked Questions
No. Summaries are pre-generated for the names in your universe from structured data — there's no conversational interface and nothing to prompt.
Context Analytics' natural-language layer derived from our quantitative feeds. A plain-English summary of why a name's sentiment moved, generated from the same structured data.
Summaries are grounded in the underlying structured sentiment and news data, not open-ended generation, every claim traces back to a scored data point.
On a rolling 24-hour window, refreshed as the underlying sentiment shifts.
Yes. The summary and the sentiment score are generated from the same engine, use the summary for day-to-day context and the score for the model.
A JSON API over REST, pairing the plain-English summary with its structured sentiment score and drivers.
Read the market in plain English.
Put natural-language context in front of your desk, grounded in the same structured sentiment data your models already trust.
company