Lead-Gen & AI Website-Proposal Automation (ProspectForge)
Overview
An internal build (“ProspectForge”) that automates lead generation and AI-written website proposals. It scrapes local-business data, evaluates prospects, and generates tailored outreach/proposal content using an LLM agent built on Firebase Genkit.
Why It Exists
Outbound agencies spend heavily on finding leads and hand-crafting proposals. This build tests whether the loop, discover businesses, qualify them, draft a personalized website proposal, can be automated end-to-end with scraping plus LLM generation.
What We Built
A Firebase-native system: Cloud Functions (TypeScript) organized into agent (Genkit flows, prompts, tools), scraping (Apify client, webhook handling, scrape status/processing), businesses, jobs, and credits modules; Firestore for persistence with security rules and indexes; Firebase Hosting for the admin dashboard. Scraping uses Apify’s Google Places crawler; generation runs through Firebase Genkit with OpenAI, including a WhatsApp agent flow. A detailed implementation plan documents phases, CI/CD, and milestones.
Technologies & Approach
Firebase Genkit was chosen for first-class Google/Firebase integration and structured AI flows; Apify provides reliable business-data scraping; OpenAI generates the proposal copy. The whole stack is serverless on Firebase Functions + Firestore, with a hosted admin UI and deploy scripting.
Outcome / Impact
A working build (≈20 commits over a focused build) proving the discover → qualify → generate pipeline. Validates Firebase Genkit as an agent framework and demonstrates the studio’s ability to assemble scraping + LLM generation into a usable automation with an admin dashboard.
Capabilities Demonstrated
- Building LLM agent flows with Firebase Genkit
- Web-scraping orchestration via Apify (Google Places)
- Serverless architecture on Firebase Functions + Firestore
- Automated, personalized content/proposal generation
- AI-assisted lead generation and qualification pipelines