AI Competitor Intelligence Report — Automatically Monitor What Your Competitors Are Shipping, Saying and Winning Every Week Without Spending
Built by AetherMatrix ·
About this build
A Claude and Perplexity powered workflow that monitors up to 5 competitors every week across their product updates, pricing changes, content output, job postings, customer reviews and social activity — then synthesises everything into a structured weekly intelligence report delivered to Notion with a 5 bullet Slack sum
What was built
Built this after spending 3 hours on a Sunday manually researching 3 competitors before a board meeting and realising I had been doing this exact same work every quarter without ever building a system around it. The information was useful. The process was completely unsustainable. The workflow runs every Monday at 6am so the report is ready before the week starts. Step 1 — Competitor profile setup in Airtable Each competitor gets a row with their company name, website, product URLs, pricing page URL, LinkedIn company page, Twitter/X handle and G2 or Trustpilot review page. Adding a new competitor to monitor takes about 2 minutes. Step 2 — Weekly research sweep via Perplexity Every Monday n8n triggers a structured research sweep for each competitor across 6 intelligence categories. Product updates — any new features, product announcements, changelog updates, beta launches or deprecated features from the last 7 days. Pricing changes — any changes to pricing pages, plan structures, trial lengths or promotional offers compared to previously logged data. Content and messaging — what topics are they publishing, what keywords are they targeting, what narrative are they pushing in their marketing. Customer sentiment — new reviews on G2, Trustpilot or Capterra from the last 7 days. Clusters of reviews mentioning the same thing are flagged as signal not noise. Job postings — what roles are they currently hiring for. Job postings reveal strategic priorities before any public announcement. A competitor hiring 5 ML engineers tells you something their blog will not say for another 6 months. Social and press — any significant social content, press coverage, podcast appearances or leadership mentions from the last 7 days. Step 3 — Intelligence synthesis via Claude All 6 research categories across all competitors get passed to Claude with a structured prompt that does four things. Significance scoring — each piece of intelligence gets rated Low, Medium or High based on how likely it is to affect the competitive landscape or require a response. Pattern identification — Claude looks across all competitor activity to identify whether multiple competitors are moving in the same direction simultaneously which often signals a market shift rather than individual company decisions. Implication analysis — for each High significance item Claude writes one paragraph on what this means for the business being monitored and what the appropriate response might be. Strategic summary — one page overview of the most important competitive developments of the week written for a founder who has 5 minutes not 2 hours. Step 4 — Report delivery to Notion The full intelligence report gets written to a dedicated Notion page automatically. Each report is dated and archived so over time there is a searchable history of competitor activity going back weeks and months. Step 5 — Slack summary A condensed 5 bullet point summary of the week’s most significant competitive developments gets posted to a Slack channel Monday morning. This is the version most people actually read — the Notion report is there for anyone who wants to go deeper. Step 6 — Airtable trend logging Key metrics get logged to Airtable weekly — pricing changes, new product features, review sentiment scores and job posting counts. After 8 to 12 weeks this data starts revealing patterns invisible week to week. What took the longest to get right: The significance scoring was inconsistent early on. Claude would rate minor blog posts as High and genuinely important pricing changes as Medium. Fixed by giving Claude explicit criteria — High means this requires a response or changes our roadmap, Medium means worth monitoring but no action needed this week, Low means log it and move on. The job posting analysis was the most valuable addition but hardest to implement. Fixed by searching for the competitor name plus site:linkedin.com/jobs in every query which significantly improved detection accuracy. The report was too long in early versions. Cut to a maximum of one page per competitor and moved supporting detail to expandable Notion sections. The constraint forces better prioritisation. Results after 10 weeks: Caught a competitor repricing their mid-tier plan by 23 percent the same week it happened rather than 6 weeks later when a customer mentioned it. Was able to prepare response messaging before the first customer asked about it. The job posting analysis flagged a competitor hiring aggressively for enterprise sales roles 3 months before they publicly announced an enterprise product tier. Had 3 months to prepare rather than being caught off guard. Customer sentiment tracking identified a cluster of negative reviews on a competitor’s mobile app — 7 reviews in 2 weeks mentioning the same crashes. Informed a product decision to prioritise our own mobile experience before we had even seen direct customer demand for it. Will share the full n8n workflow JSON and Claude synthesis prompt in comments for anyone who wants to deploy this.