Build a Branded Competitor-Analysis Report Engine: Auto-Discover, SWOT, and Ship a Branded PDF (Productized-Service Teardown)

The problem

Founders, marketers, and agencies want ongoing competitive intelligence — who the real competitors are, where the positioning gaps sit, and a SWOT they can act on — but the work behind it is brittle. Doing it by hand means opening tabs, reading competitor sites and pricing, scanning reviews, and stitching it into something presentable, which practitioners describe as roughly 6-8 hours a month per account. Rebuilding that as a fragile n8n flow tends to break, and the off-the-shelf monitoring platforms either start at hundreds of dollars a month behind a sales call or hand you a dashboard, not a deliverable you can put your own brand on and sell. The specific gap is producing a branded, client-ready report on a repeatable cadence without the cost and inconsistency creeping back in every cycle.

Who it's for

Founders, marketers, SaaS teams, and especially agencies that want to sell competitive intelligence as a recurring, productized service. It fits anyone who needs a branded, client-ready competitor report on a monthly or quarterly cadence — and who would rather own the engine that produces it than rent a per-seat monitoring platform or assemble each report by hand.

How it works

1. 1

   Onboard the agent with context: write a CLAUDE.md (or a WAT-framework doc) describing the business, the analysis dimensions you care about, and the report cadence, then drop in brand assets so the agent can extract the logo, color palette, and typography it will brand the PDF with.

2. 2

   Spec the deliverable in plan mode: tell the agent what the output is — auto-discover the competitor set from a seed list, the SWOT and gap dimensions to score, the per-competitor breakdown, and how often it re-runs — before it writes any code.

3. 3

   Have the agent build the tool chain as discrete steps: collect_business_info (your positioning), discover_competitors (expand the seed set into the real field), research_competitors (crawl offerings, pricing, and positioning), analyze (gap + SWOT scoring across the dimensions), and generate_pdf (the branded output).

4. 4

   Wire the research tools: a crawl/extract tool (Firecrawl) to pull competitor pages cleanly and a research API (Perplexity) for breadth, with the analysis reasoning handled by the Anthropic API (Sonnet).

5. 5

   Cache the researched competitor data to disk so each monthly re-run only refreshes what changed instead of re-crawling everything — this is what keeps recurring runs cheap and the output consistent run-to-run.

6. 6

   Generate the branded PDF with ReportLab and matplotlib: a multi-page report carrying your logo and colors, charts for the gap/SWOT view, plus a separate file per competitor, ready to send to a client as the deliverable.

Tools

The result

You get an engine that turns a business and a seed competitor list into a branded, client-ready report on a repeatable cadence instead of a 6-8-hour manual research grind each cycle. The mechanism is a discrete tool chain: the agent auto-discovers the real competitive field from your seed set, researches each competitor's offerings, pricing, and positioning, runs a structured gap and SWOT analysis across the dimensions you defined, and renders a multi-page PDF in your logo, colors, and charts — plus a separate file per competitor. Because the researched data is cached to disk, each monthly or quarterly re-run only refreshes what actually changed, which keeps the per-cycle cost low and the output format consistent run-to-run — the property that makes a fixed retainer (rather than a one-off project fee) economically sane to offer. Honest framing: the discovery and analysis are AI-assisted research a human should still sanity-check before it goes to a client, and the value is owning a branded, recurring deliverable engine rather than renting a per-seat monitoring dashboard — no specific time-saved or accuracy figure is guaranteed.