ForgePulse

Checking booking flow, trust signals, accessibility and local visibility by hand repeats the same work for every business. An operator needs a structured review and prioritised next steps, with enough evidence to question the generated advice.

Bywater built an operator-led product demonstration: business intake, website scanning, scoring, competitor context, report generation and tokenised sharing. It is a reusable audit workflow, with synthetic seeded businesses for portfolio review.

The workflow combines deterministic scan signals with AI-generated explanation. Provider fallback and deterministic demo output keep a missing model key from breaking the demonstration. The repository separates architecture, test commands and production boundaries, giving AI-assisted changes a concrete scope to verify instead of treating generated reports as automatically correct.

SQLite keeps local demos simple; the documented production path calls for Postgres. The scanner treats submitted URLs as untrusted and includes SSRF checks for private addresses and redirects. Reports use bearer-token links that can be expired or regenerated, while admin writes check their origin.

The source documents the complete seeded scan-to-report demonstration, and the existing screenshot shows its shareable report. This review verified the repository documentation and demo asset, not a new live scan. No claim is made that a business grew revenue, leads or search ranking because of the tool.

Scan work currently runs in an in-process background promise; durable jobs, shared rate limits and monitoring are needed before paid or high-volume use. It has no billing or self-service onboarding. Email delivery is intentionally disabled in the portfolio build.