FlowRunner
PricingContact
Theme
Start Free

finlight

Analytics & Data

finlight is a financial news API with sentiment analysis and resolved company entities. Agents search articles by company, ticker, or keyword, look up a story by URL, and browse the catalog of news sources to feed research and alerting flows.

3 actions API key available
finlight website Platform Documentation Capability data verified 2026-08-25
A scheduled screen runs against the counterparty watchlist
Search Articles runs one advanced query per name, ordered by indexed date
Get Article By Link pulls in any story a colleague forwarded, in the same shape as the screen results
List Sources records which publications were actually in scope and which were indexed but excluded
The screen posts to the analyst with the sentiment, the resolved tickers and the coverage footprint attached
The analyst decides whether the coverage that ran is wide enough to support the credit decision waiting on it

What This Integration Enables

finlight is for teams who need a screen rather than a feed. The difference matters. A feed hands you everything and leaves the filtering to a person, which means the filtering does not happen. A screen encodes the condition you actually care about and returns only what meets it. finlight's query field takes an advanced syntax, so a real screen is a single request: a company name, combined with the events that would change your view, minus the noise you already know you do not want. Results come back with sentiment attached, and on plans that include it, with the companies each article mentions resolved to tickers and ISINs, which is what lets an agent join a story to a position, a customer, or a supplier on your own books.

Three properties of the API decide whether a monitoring flow is honest, and FlowRunner's connectors surface all three rather than hiding them. Language defaults to English and silently excludes everything else, so a flow watching a European supplier sets it deliberately. Ordering by indexed date rather than publish date is what makes polling correct, because some sources publish with backdated timestamps and a cursor walking publish dates would step straight over them. And a source being indexed does not mean it reaches your results, because a source without the default flag is excluded unless you name it. Those three facts are the difference between a screen that found nothing and a screen that was never looking.

Without FlowRunner

A feed nobody reads Headlines arrive all day, and the ones that matter are indistinguishable from the ones that do not
Names that do not resolve Articles mention companies in prose, and matching them to the entity on the ledger is manual work
Coverage assumed, never stated A monitoring report says nothing was found without saying what it looked at

With FlowRunner

A screen, not a stream One advanced query per counterparty returns the stories that match the condition being watched
Companies resolved to identifiers Articles arrive with the companies they mention resolved to tickers and ISINs on plans that carry it
Coverage stated with the result Every screen reports the language filter and the source set it ran under, so silence can be read honestly

Use Case Scenarios

  • A standing screen on the counterparty book

    The customer and supplier list is the watchlist. On a schedule, the agent runs Search Articles once per name with an advanced query that pairs the company with the events that would change a credit view, ordered by indexed date so nothing backdated slips past the cursor. Hits are grouped by counterparty, tagged with sentiment, and posted to the credit review channel in Slack with the article link and the resolved tickers. Clean names produce no message at all, so the channel carries findings rather than volume.

  • Enriching a link somebody already sent you

    Someone forwards a story, or a webhook from another system carries a URL. Rather than searching for an article the flow already has, the agent calls Get Article By Link, which returns that exact piece with its sentiment and, optionally, the companies it mentions. The enriched record appends to a research log in Notion alongside the analyst's own note. A 404 from this route means finlight never indexed that URL, which the flow records as a coverage gap rather than as a failed request.

  • Auditing what the monitor can actually see

    Once a quarter, the agent calls List Sources and writes the full index to Google Sheets: every source finlight carries, its domain, its origin country, its languages, and whether it is served by default. The team reads down the sheet for the trade press and regional outlets they consider load bearing, and any source that is indexed but not default gets added to the opt-in list on the screens that need it. The vendor scrapes on a best effort basis and states the list may change, which is exactly why this runs on a schedule rather than once at setup.

Human-in-Loop Highlight

The dangerous output of a news monitor is not a bad headline. It is a clean report. When a credit limit increase, a contract renewal or a payment release is waiting on "no adverse coverage found", the absence of results is being read as evidence, and on this API an absence has several innocent explanations that look identical from the outside. The language filter defaulted to English and excluded the counterparty's home press. The publication that would have carried the story is indexed but not a default source, so it never entered the result set. finlight simply has not indexed that outlet. None of those produce an error. All of them produce silence. So when a screen comes back empty on a name that has a decision riding on it, the agent does not clear the counterparty. It posts to the analyst: "Screen on Meridian Logistics returned no adverse coverage. It ran in English only, against the default source set, over the last ninety days. Their home market press is not in that set. Treat this as cleared, or should I re-run with the language filter lifted and the regional sources opted in before the limit increase goes through?" The analyst answers, and the flow records which coverage the decision was actually made on. An agent can search faster than any analyst. Deciding that a silence is good enough to lend against is a judgment, and this is where FlowRunner's human-in-the-loop design puts it.

Agent processes routinely
Detects exception requiring judgment
Clear match Continues automatically
Ambiguous Routes to human via preferred channel
Human decides
Agent resumes with decision

Agent Capabilities

3 actions

Articles

2
  • Search Articles Searches finlight's financial news index and returns articles with their sentiment and, on a plan that includes it, the companies each one mentions with resolved tickers and ISINs. The query field takes finlight's advanced syntax, so a real screen is one request. Tickers matches companies named in the article, and Countries matches what the article is about rather than where the publisher sits. Language defaults to English and silently excludes everything else, and ordering by indexed date rather than publish date is what makes a polling flow correct.
  • Get Article By Link Fetches a single article by its URL, with sentiment and optionally the companies it mentions. Built for enriching a link you already have, from a webhook, a feed or a customer message, rather than searching for it. A 404 means finlight never indexed that URL, which is a coverage gap rather than a mistake in the request.

Sources

1
  • List Sources Lists every news source finlight currently indexes, with its domain, origin country, languages and whether it is served by default. The default flag is the field that matters: a source without it is indexed but not included in results unless it is named in the opt-in list, which is the usual reason an expected publication is missing from a search. The vendor scrapes on a best effort basis and states the list may change.

Frequently Asked Questions

What can FlowRunner do with finlight?

FlowRunner agents can run Search Articles, Get Article By Link, and List Sources in finlight.

Does connecting finlight to FlowRunner require OAuth?

No. finlight connects to FlowRunner with an API key, no OAuth flow required.

Can finlight trigger a FlowRunner workflow automatically?

finlight doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.

Start building with finlight

Free plan, no card required. Connect in minutes.