Product Overview
LeadFella is an AI sales assistant for agencies, software and web-development firms, SEO agencies, and freelancers. It discovers local businesses, analyzes their digital presence, finds the work they likely need, ranks them by revenue potential, drafts personalized outreach, and manages the whole sales pipeline in a built-in CRM.
It is one standalone web product: a React single-page app served by a single Java Spring Boot backend (a modular monolith) that does everything - including Google Maps scraping via Playwright-Java. Scraping is one capability among many, not the whole product.
Who it is for
- Agencies (web, design, marketing) looking for local businesses to pitch.
- Web-development firms that win work by spotting outdated or missing sites.
- SEO agencies that lead with a concrete, observed audit.
- Freelancers who need a single tool from prospecting through to close.
What it does, end to end
- Find local businesses from Google Maps - by hand or by describing a niche, a location, and how many leads you want.
- Analyze each lead's website: contacts, technology stack, SEO, quality and maturity scores, and the opportunities the business likely needs.
- Prioritize by revenue potential and expected value (project value weighted by close probability), so the list sorts by money on the table.
- Personalize outreach with AI grounded in the real website analysis.
- Reach out over tracked email - always Draft → Review → Approve → Send.
- Manage the relationship in a Kanban CRM with deals, tasks, notes, and a full activity timeline.
- Learn from a funnel/conversion analytics layer that sharpens targeting over time.
See features.md for the full capability reference.
Core principles
- Modular by hard boundary. Each capability is a separate feature package inside the backend with a clear internal API. Packages do not reach into each other's internals.
- One website, one backend. A single React SPA talks only to the Spring Boot
backend over
/api. One database (MongoDB). One command runs it (no Docker). - Revenue first. The minimum path to sending real outreach is built before any "build later" feature.
- Never auto-send. Every outreach message follows Draft → Review → Approve → Send. LeadFella never sends without explicit approval.
- Deterministic before AI. Scoring and opportunity logic are rule-based and explainable; LLMs layer on top for personalization only.
Capability phases
| Phase | Capability | Summary |
|---|---|---|
| 1 | Lead Collection | Google Maps scraper + AI scrape + lead database |
| 2 | Website Intelligence | Crawl, contacts, tech detection, SEO, scores, opportunities |
| 3 | Revenue Opportunity | Revenue potential, project value, expected value |
| 4 | AI Personalization | Lead summary, sales angles, email/WhatsApp copy |
| 5 | Outreach | Tracked email with Draft → Approve → Send |
| 6 | CRM | Kanban pipeline, deals, tasks, notes, timeline |
| 7 | Analytics | Funnel, conversion, revenue, learning loop |
The product also ships an in-app Docs section, Bearer-token authentication, settings (including AI/vision configuration), and multi-currency support (defaults to INR).
The product vs. this website
This documentation set lives in the marketing website project, which is a separate codebase from the product app:
- Product app - the React SPA + Spring Boot backend described above
(
../../apps/weband../../backend). Served in production atapp.leadfella.com. - Marketing website - the public site, early-access waitlist, contact form,
and admin panel documented here (
../frontendand../backend). Served atleadfella.com, with its small API atapi.leadfella.com.
For how the two relate, see architecture.md.