Key points
- Perplexity launched a Monitor API that runs scheduled web searches or page checks and pushes only relevant changes to agents through signed webhooks.
- Pricing is $1 per 1,000 successful web monitor runs or per 1,000 fetched URLs for page monitors; it is live now and uses the existing Perplexity API key.
- In its own early test of 60 topics, Perplexity reported 63 fresh, relevant pages in eight hours, versus 20 for Firecrawl and 15 for Exa.
- Schedules run from every 15 minutes to weekly, and page monitors can watch up to 10 pages.
Why it matters for marketers: Marketers can feed competitor pricing, funding and regulatory changes to agents without building their own polling and deduplication code, though the performance claims come from Perplexity's own testing.
Perplexity has released a Monitor API that keeps watch on the web for developers and sends updates to their agents by webhook. Until now, teams had to schedule their own repeat searches and write code to strip out duplicates and spot changes. Perplexity says that burden becomes unworkable at scale.
The tool offers two approaches. One follows a search topic over time, such as funding news. The other tracks specific pages like pricing or policy pages, saving a baseline and then returning a text diff when something shifts. Developers describe their goal in plain language, and a review model checks relevance, source credibility, novelty and timeliness before an alert goes out, with a confidence score attached.
For marketers and content teams, this is a practical way to automate competitive and market watching. Perplexity itself points to agents that flag competitor price changes, track target accounts for sales, or summarize new rules for compliance staff. An agent that only wakes when something meaningful happens can cut noise compared with a daily digest.
Treat the benchmark with caution. Perplexity compared itself with Firecrawl, Exa and Parallel across 60 topics and called the numbers early internal results, with a separate model doing the scoring. An independent test would be more telling. Teams considering it should pilot a few monitors on competitor pages, tune the thresholds, and check how many alerts are actually useful before expanding.
Sources
This brief is our own summary. Read the full reporting at the source.