Hot AI News

Perplexity’s Monitor API watches the web for you, from $1 per 1,000 runs

Perplexity's Monitor API watches a topic or up to 10 pages on a schedule and sends a webhook when something relevant changes, at $1 per 1,000 successful web runs.

Perplexity’s Monitor API watches the web for you, from $1 per 1,000 runs
Image: Perplexity

Perplexity has launched the Monitor API, a service that watches the web or specific pages on a schedule and sends a webhook when something relevant changes. According to the official announcement, web monitors cost $1 per 1,000 successful runs.

The pitch is aimed at developers who build agents that need to react to new information. Perplexity says most teams currently run scheduled searches and keep custom code to remove duplicates and spot changes, an approach that gets harder across hundreds of topics and pages.

Figure from Perplexity's Monitor API announcement showing how updates are checked and delivered
Image: Perplexity

Two monitor types, one filter in between

A monitor is defined with a search query or a list of pages, plus an optional goal written in plain language. Perplexity’s example asks for new funding rounds in AI infrastructure, with the goal of surfacing Series A or B announcements. The goal can be edited later without recreating the monitor.

There are two modes. A web monitor runs its query through Perplexity Search across web and news results to find new developments on a topic, company or question. A page monitor follows up to 10 pages you name, for instance a pricing page, a changelog or a policy page. The first check stores a baseline, and later runs return a text diff of what changed.

Monitor API illustration with orange light beams and dotted patterns on a dark background
Image: Perplexity

Between the check and the alert sits what Perplexity calls Decider, which tests every result against the goal. A web result has to clear four tests: it must be relevant, come from a legitimate source, carry information that is new, and be recent. For page changes, Decider judges whether the change is meaningful for the goal. An update goes out only if every test clears its threshold, and each one carries a confidence score. The thresholds can be adjusted per check.

Advertisements

Schedules, signed webhooks and per-run billing

Schedules can be a repeating interval, from every quarter hour up to weekly, or a custom cron expression. Every run resumes at the point where its predecessor finished, and Monitor filters out duplicates across runs. Delivery is by signed webhook, with automatic retries when a delivery fails for a transient reason. A monitor can also notify its owner when it pauses itself, and can be paused, resumed or triggered on demand from the API or the API Console.

Billing differs by mode. A web monitor costs $1 per 1,000 successful runs, while a page monitor costs $1 per 1,000 URLs fetched, because one run can fetch many pages. A run is one scheduled or on-demand check. Perplexity says the API is available today and uses the same API key as its Search, Context and Agent APIs. Creating a monitor takes one POST request to https://api.perplexity.ai/v1/monitors with a name, targets, a goal, a schedule and a webhook URL. Metadata can be attached so each update is routed to the right user, workspace or agent.

The company lists four use cases: go-to-market teams tracking funding, executive hires and launches; equity research on earnings, guidance and adverse media; competitive intelligence on pricing and changelogs; and regulatory monitoring routed to compliance teams.

Where Monitor fits in Perplexity’s agent push

Monitor arrives in the middle of a run of agent-oriented launches. On November 5, Perplexity introduced Teammates that work on their own and can message coworkers in Slack. On November 14, those agents reached the Mac app for Max subscribers only. On the same day as this launch, we reported that the company is planning an automated research effort backed by more than 10,000 GPUs.

Those products are consumer and prosumer agents. Monitor is infrastructure: an event source that other people’s agents can subscribe to. Its webhook model matches the trigger pattern the company describes, where an alert starts a workflow such as researching an account after a funding announcement. It also reuses the Search API key and billing relationship, which suggests Perplexity wants developers to treat search, context, agents and monitoring as one platform. That reading is our inference, not something the announcement states.

Advertisements

What the announcement does not establish

The performance claim rests on Perplexity’s own test. It ran Monitor against other Search API providers on 60 topics taken from production Search API queries, covering company news, technology releases and regulation. Each monitor watched its topic for eight hours, and a separate model then graded relevance and publication timing. Perplexity says it delivered the most fresh, relevant results across the widest range of topics, and that many of its results were not returned by any other provider.

The post does not name the competitors, the grading model or the scores, and the comparison is shown only in a figure. An eight-hour window across 60 topics is also a narrow sample for a service meant to run for weeks. Buyers will want to test their own topics.

Cost is the other open question. A web monitor checking hourly makes 24 runs a day, or about 720 over a 30-day month, which at the stated rate is a fraction of a dollar. Pricing therefore looks modest for each monitor, but a team with hundreds of topics or 15-minute schedules would see volume add up. The announcement does not say whether failed runs are free beyond the phrase “successful runs”, nor whether any rate limits or caps on the number of monitors apply. Regional availability is not mentioned either.

Signs to watch are whether independent developers confirm the freshness claim, whether the confidence scores prove calibrated enough to trust without a human check, and whether Perplexity publishes the evaluation details. The announcement points to the Monitor API docs and the API Console for anyone who wants to try it.

Sign Up for the Daily AI Pulse

One email a day. All the stories that matter.

Alex CarterAI News & Big Tech Correspondent

Alex Carter writes for Aiholics, keeping readers updated on the fast-paced world of AI and Big Tech.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may also like