TL;DR: A competitor monitoring agent built from Perplexity, Make, and Slack surfaces launches, pricing shifts, and hiring signals inside 10 minutes of them going public. Setup takes 45 minutes. It runs for under $20 a month.
You don’t need a competitor monitoring agent because your rivals are scary. You need one because you have four tabs open, three clients waiting, and no time to babysit ProductHunt at 2am. The build below catches the signals that actually matter and dumps them into the same Slack you already stare at.
What you’ll get in this tutorial
- A working Perplexity + Slack agent that runs on autopilot
- The exact prompt template I use to filter noise from real signals
- A full cost breakdown under $20 per month
- The tuning rules that keep alert fatigue at zero after week one
Why a competitor monitoring agent beats manual tracking
Manual tracking means Google Alerts for a brand name, LinkedIn scrolling, and hoping. Google Alerts misses paywalled news, most SaaS blog posts, and everything on X. LinkedIn shows you what its algorithm wants you to see, which is rarely a competitor’s new pricing page.
A competitor monitoring agent using Perplexity fixes three problems. Perplexity’s Sonar model searches the live web with citations by default [source-needed]. You can write one prompt template and reuse it across every competitor. And you get structured JSON output you can filter, so Slack only pings you when a signal beats your confidence threshold.
I ran this exact stack against 5 SaaS competitors in my niche for 3 weeks in June [test-claim]. Slack fired 34 alerts. Six were worth acting on. That’s a 17% signal rate. Sounds low until you compare it to scrolling LinkedIn at midnight and remembering none of what you read.
The stack behind the competitor monitoring agent: Perplexity + Make + Slack
You need four pieces and nothing more:
- Perplexity API access, $5 minimum credit purchase [verify pricing]
- Make.com for scheduling and routing, free tier for up to 10 competitors [verify pricing]
- A Slack workspace with a dedicated
#competitorschannel - A Notion database or Google Sheet as the alert log
Why Perplexity over ChatGPT with browsing? The Sonar API is cheaper per call and returns citations in the response body by default [source-needed]. ChatGPT’s web tool isn’t exposed for stable scheduled use, so you’d be stapling extra pieces on to fake it.
Why Make over Zapier? For scheduled AI + webhook work, Make gives you more operations per dollar {{internal:make-com-vs-zapier-comparison}}. If you already pay for Zapier, don’t add a second tool. Use Zapier’s Schedule + Webhooks combo. The logic is identical.
Step 1 — Pick 5 competitors and 5 signals for your competitor monitoring agent
Don’t monitor 20 competitors. You’ll drown and mute the channel by Thursday. Pick 5:
- 2 direct competitors — same feature set, same buyer
- 2 adjacent competitors — different feature, same buyer
- 1 aspirational — 10x your size, so you can copy their playbook
Then pick 5 signals per competitor:
- Pricing page changes
- Product launches or major feature announcements
- Job posts, especially AI, growth, or sales roles
- Funding rounds
- Blog posts or podcast appearances by the founder
Log these in {{internal:tracking-competitors-notion-template}} before you build anything else. If you can’t name the 5 things you care about, no agent will save you from your own indecision.
Step 2 — Build the Perplexity query template
This is the query I drop into Make’s Perplexity API request module:
In the last 7 days, has [COMPETITOR_NAME] done any of the following:
1. Changed pricing on [PRICING_URL]
2. Launched a new product or major feature
3. Posted 3 or more new open roles on LinkedIn or their careers page
4. Announced a funding round
5. Published a blog post or podcast featuring their founder
Return a JSON array. For each signal found, include:
- signal_type: one of ["pricing","launch","hiring","funding","content"]
- confidence: 0 to 1, where 1 means directly cited from a primary source
- summary: 2 sentences max
- source_url: primary source URL
- action_for_us: 1 sentence recommendation for a competing solo SaaS founder
If no signal is found for a category, omit it. If nothing is found at all, return [].
Set the model to the current top Sonar tier [verify pricing]. Temperature 0.1. Low enough to keep JSON parseable and stops the model from inventing hiring pushes that never happened.
Step 3 — Wire the competitor monitoring agent to Slack
Inside Make, the scenario looks like this:
- Schedule trigger, every 12 hours
- Iterator over your 5 competitors, sourced from Notion or Sheets
- Perplexity module running the template above, with the competitor name and URLs interpolated in
- JSON parser on the response
- Filter — only continue if
confidence >= 0.7 - Slack module — post to
#competitorsusing the format below
The Slack message format that stopped me from muting the channel:
🚨 [Competitor Name] — [signal_type]
[summary]
Confidence: [confidence]
Source: [source_url]
👉 [action_for_us]
Also append every alert, including sub-0.7 confidence ones, to a Notion Competitor Log. You want the filtered ones visible somewhere so you can review calibration on Friday. If you skip this step, you’re flying blind on whether your threshold is right.
Step 4 — Tune the agent so Slack doesn’t turn into noise
Week one, you will get too many alerts. That’s expected. Here’s how you fix it without ripping the whole thing out.
- Raise the confidence threshold from 0.7 to 0.8 if 40% or more of your alerts are noise
- Add a de-dupe step — if the same
source_urlwas posted in the last 7 days, skip it - Add a keyword blocklist for daily publishers (weekly digests, episode counts, changelog posts)
- Drop the schedule from every 12 hours to every 24 hours if your niche moves slowly
Every Friday, open {{internal:weekly-competitor-review-checklist}} and score last week’s alerts. Anything you acted on or would have acted on is real signal. Anything else is noise. Recalibrate the query with the noise patterns you spotted. Ten minutes a week keeps this thing honest.
What the competitor monitoring agent costs to run
Realistic monthly cost for 5 competitors, checked every 12 hours:
- Perplexity API: roughly $8 to $15 [verify pricing]
- Make.com: free tier, since 1,000 operations per month covers this easily [verify pricing]
- Slack: free tier
- Notion: free tier
Total: under $20 a month if you stay disciplined about the 5-competitor cap. That’s cheaper than a single hour of a virtual assistant doing this badly. Enterprise competitive intelligence platforms start closer to $500 to $1,000 a month and are built for marketing teams of 10 plus [source-needed]. For a solo founder, that math never lands.
Bottom-line recommendation
Build the Perplexity + Make + Slack version. Skip the enterprise competitive intelligence platforms until you have a marketing team of three or more, and even then think twice.
Perplexity handles the searching. Make handles the plumbing. Slack handles the notification. Notion holds the log. Start with a confidence threshold of 0.7, cap yourself at 5 competitors, run every 12 hours, and review every Friday for 10 minutes.
If you already run Zapier heavily, swap Make for Zapier. Don’t add a second automation tool for a single workflow. That’s how you end up paying twice for the same job.
FAQ
Do I need to code to build this?
No. Make wires everything visually. The only code involved is the JSON prompt template, and that’s copy-paste.
What if Perplexity hallucinates a signal?
The 0.7 confidence filter plus the required source_url catch most of it. Every Slack alert links to a primary source, so a 30-second click verifies whether it’s real before you act.
Can I use ChatGPT or Claude instead of Perplexity?
Only Perplexity’s Sonar API returns real-time web results with inline citations natively [source-needed]. Claude and GPT-4 need extra web-search tooling bolted on. If you already pay for one of them, Perplexity is still the right call for this specific job.
How many competitors is too many?
More than 8 in one channel and your brain starts filtering the channel out entirely. If you need more coverage, split into #competitors-direct and #competitors-adjacent.
Does this work for agencies or local businesses, not just SaaS?
Yes. Swap pricing page for new service page and product launch for new case study. The stack doesn’t change.
Will competitors know I’m watching them?
Perplexity reads public data. No login, no scraping detection, no tripped alarms. You’re reading their marketing exactly how they want it read.
What to do in the next 10 minutes
- Open a sheet and list your 5 competitors with their pricing, blog, and careers URLs. 5 minutes.
- Buy $5 of Perplexity API credit and copy the API key into a password manager. 2 minutes.
- Paste the query template into a Make scenario and run it manually against your top competitor. If the JSON parses, wire up the Slack step next. 15 minutes to first alert.
Twenty-five minutes from now, you’ll have a competitor monitoring agent shipping real signals into Slack while you go back to your actual job.