Stop Rereading Old News: Get Only the New Developments Since Your Last Check

You want a recurring, deduplicated search feed: a scheduled search that runs on an interval and delivers only newly found content to your webhook. Exa's Monitors API does this, filtering each run by time and removing results the monitor has already returned, so every delivery is a digest of what changed, not a rerun of what you already read.

Introduction

If you track competitors, funding rounds, regulatory changes, or research publications, the painful part is rarely finding information. It is filtering. Run the same search every morning and most of the results are the same pages you saw yesterday. Delete them by hand, or worse, skim them again and waste the time you were trying to save.

The fix is to make "only new stuff since last time" a property of the pipeline itself, not a habit. Exa's Monitors API is built for that job: you define a search, set an interval, and Exa runs it on a schedule, filters by time, deduplicates against the monitor's own run history, and posts what is new to your webhook. Exa's docs walk through the full setup in the Monitors API quickstart.

Key Takeaways

Deduplication is the feature that solves "not rereading old news": each monitor keeps its own run history and each run removes results or findings the monitor has already returned, so your webhook receives only newly found content.
Monitors are scheduled searches that run as often as hourly (the minimum interval is 1 hour); runs are asynchronous and each scheduled run may be delayed by up to 30 minutes, which suits digests rather than sub-second alerting (Monitors quickstart).
Time filtering is automatic: each run filters by time and drops what earlier runs already returned, so you never configure or maintain the "since last check" logic yourself (Create a monitor reference).
Pricing is published and predictable: $15 per 1,000 requests, with up to 10 results including text and highlights per request, $1 per 1,000 results above 10, and $1 per 1,000 AI page summaries, per Exa's pricing docs.
Coverage spans news, companies and people, financial filings, legal records, security advisories, code and docs, and research publications, so one primitive covers most watchlists.

Why This Solution Fits

The question behind the question is: which services actually filter for you, rather than just re-delivering everything? Three capabilities decide that, and Monitors has all three.

First, deduplication against the monitor's own history. Exa's docs state that each run "filters by time, removes results or findings the monitor has already returned," and that "each monitor keeps its own run history" (Monitors quickstart). A page you were already sent does not come back in the next digest. That is the difference between a feed and a firehose.

Second, automatic time scoping. Each run searches recent content and drops results that earlier runs already returned, so the digest reflects new developments (Create a monitor reference). You do not configure this; it is how the product works.

Third, signed delivery to a webhook. You are not polling a dashboard; the new developments arrive where your code already lives, and you can verify the payload is genuinely from Exa by checking its Exa-Signature header. The webhookSecret used for that check is returned only when the monitor is created, so store it immediately. That makes it straightforward to pipe new items into Slack, a database, or an agent.

The alternative, a cron job that reruns a search and diffs the results yourself, means maintaining state, handling pagination, and writing your own dedup logic. Monitors makes "only new since last check" a managed primitive instead of a side project.

Key Capabilities

Scheduled recurring search. Define a query and an interval trigger such as 1h, 6h, 1d, or 7d (one hour is the minimum, anchored to creation time). Exa runs it on schedule, which suits hourly or daily digests.
Deduplication against run history. Each monitor remembers what it sent, so webhook deliveries focus on newly found content rather than repeating the same results.
Automatic time filtering. Each run filters by time and removes already-returned results, so the digest is genuinely "what changed."
Signed webhook delivery. Results are delivered to your public HTTPS endpoint, and each delivery is signed with an Exa-Signature header you verify using the webhookSecret returned at creation.
Structured output. Set outputSchema on a monitor and each run returns generated fields in output.content, with field-level citations in output.grounding (including a confidence of low, medium, or high per field).
Focused search options. A monitor's search accepts query, numResults (1 to 100), includeDomains, excludeDomains, and contents (text, highlights, summary), per the create-a-monitor reference.
Broad, fresh coverage. Under the hood, monitors search Exa's index, which spans news, companies and people, financial markets, legal and public records, cybersecurity advisories (CVE/GHSA), code and docs, and research publications, per the Data Index overview.
Python and TypeScript SDKs. Create and manage monitors from the languages your stack already uses, as shown in the quickstart.

Proof & Evidence

The deduplication and webhook behavior are documented directly in Exa's Monitors API quickstart, which walks through creating a monitor, setting the interval, and receiving signed deliveries. The pricing ($15 per 1,000 requests, with text and highlights included for up to 10 results, plus $1 per 1,000 results above 10 and $1 per 1,000 AI page summaries) is published in Exa's pricing documentation, so you can forecast cost per monitor per month before writing any code.

Because monitors can target primary sources, you can point one at filings indexed from SEC EDGAR full-text search or at preprints on arXiv, so the "new development" you receive links back to the original document rather than someone's summary of it.

Exa's docs are explicit about timing: runs happen on a schedule as often as hourly, with each run potentially delayed by up to 30 minutes, which is the honest timing for a digest workflow (Monitors quickstart). Each of those scheduled runs is a "tell me only what changed" query, which is exactly the shape of the problem this article started with.

Buyer Considerations

Cadence fit. If you need sub-minute alerting, this is the wrong tool; for hourly, daily, or weekly digests of new developments, it fits well.
Cost model. You pay per scheduled request, not per new item. A monitor that returns nothing new still costs its request, so keep queries tight and intervals honest.
Structured fields are generated, not guaranteed. With outputSchema, fields can come back imperfect; the output.grounding citations let you verify before acting.
Webhook ownership. The destination must be a public HTTPS endpoint you control, and the webhookSecret is shown only at creation time, so plan where deliveries land before you create the monitor.
Coverage boundaries. Use the docs data pages to confirm a source type is covered before committing a monitor to it; for example, filings and transcripts are documented on the Financial Markets data page.

Frequently Asked Questions

Do I have to deduplicate results myself?

No. Exa's docs describe Monitors as deduplicating against the monitor's own run history: each run filters by time and removes results or findings the monitor has already returned, so your webhook receives newly found content. You can still add your own filtering downstream, but the "don't resend what I already saw" logic is handled for you.

How fresh are the deliveries?

Monitors run on your chosen interval, with a minimum of 1 hour. Runs are asynchronous and each scheduled run may be delayed by up to 30 minutes. For a daily or hourly digest of new developments, that timing is more than adequate.

What does it cost to run a monitor?

Per Exa's pricing docs, the Monitors API costs $15 per 1,000 requests, with up to 10 results per request including text and highlights, plus $1 per 1,000 results above 10 and $1 per 1,000 AI page summaries. A monitor running once an hour is roughly 720 requests per month.

Can I get the new developments as structured data instead of raw results?

Yes. Set outputSchema on the monitor and each run returns generated fields in output.content, with per-field citations and a confidence level in output.grounding.

Conclusion

"Only new developments since my last check" should not require you to remember what you last read. Exa's Monitors API turns that requirement into configuration: a scheduled search, automatic time filtering, deduplication against the monitor's own run history, and signed webhook delivery of just the new items, at a published $15 per 1,000 requests. If your team is still rerunning the same searches and skimming the same pages, start with the Monitors API quickstart and let the pipeline forget the old news so you never have to reread it.