Back to Work
Next.js + Firebase + GeminiAIWebStrategy

Media & Consumer VC

An AI deal-sourcing platform that runs itself

An investment team was sourcing companies by hand: Googling, copy-pasting into spreadsheets, and still missing most of the market. I designed and built two internal tools that find, enrich, score, and track companies automatically. The team now runs around 1,000 companies a week through a live pipeline instead of burning 10 to 20 hours on manual research.

Enter a company name and website. AI enrichment handles the rest: description, social links, follower counts, sourcing grades, and more.

Scope

Sourcing platform for investment deal flow

Retail intelligence hub

AI enrichment pipeline design

Real-time collaborative data infrastructure

Tech Stack

Next.js 15React 19TypeScriptTailwind CSSShadCN UIFirebase FirestoreFirebase AuthFirebase StorageFirebase Cloud FunctionsGoogle GenkitGemini AIBright DataPerplexity APIGoogle Maps APIGoogle PSERecharts

The Challenge

The client is an early-stage VC firm focused on media and consumer: music, entertainment, film, digital media, beauty, hospitality, and more. As a lean team without enterprise tooling, researching a potential investment meant hours of manual Googling, scattered notes, and no structured way to enrich, grade, or track companies at scale. The work cost 10 to 20 hours a week and still missed most of the market.

The Approach

01

Started with the real bottleneck: every company had to be researched by hand. I built a sourcing platform where the team enters a name and a website, and the system does the rest.

02

Designed the enrichment pipeline in phases: domain discovery, web scraping, AI analysis, social data extraction, and follower counts. Each phase validates against multiple sources, so nothing rests on a single data point.

03

Built the tools the team needed around it: theme-based organization with drag-and-drop, a grading system, bulk CSV imports, and a real-time dashboard tracking ROI and pipeline health.

04

Added a second tool for retail intelligence. Analysts photograph store shelves, AI identifies every brand and SKU on sight, then enriches each with manufacturer data and plots it on a map.

Key Features

Multi-phase AI enrichment: domain discovery, web scraping, AI analysis, social data, consensus scoring

Four concurrent background processing queues for parallel enrichment

Hierarchical theme and folder management with drag-and-drop organization

Interactive dashboard with ROI tracking and sourcing grade distribution

Brand extractor for bulk CSV imports with automated domain discovery

Store-walk photo analysis: AI detects every brand and SKU on retail shelves

Two-pass retail enrichment with Google Search grounding for manufacturer data

Interactive Google Maps view with intelligent markers per store location

A/B/C/D company grading system with pipeline tracking

Real-time Firestore listeners, so every user sees data updates instantly

The Results

~0

Companies / Week

0+

Hours Eliminated / Year

0+

Companies Processed

$0

Cost / 1K Companies

Around 1,000 companies enriched per week at $10 to $15 in API costs, work that would take a full team more than 250 hours.

13,000+ hours of manual research eliminated per year, roughly the output of six full-time analysts.

8,800+ companies sourced, enriched, and graded in the system to date.

The deal pipeline went from scattered notes and spreadsheets to a structured, searchable platform the whole team uses daily.

Enriched company data now shows up in real time. Research that used to take hours is ready in seconds.

This project pulled in

AIWebStrategy

Research, software, and design, handled by one person from scope to ship.

See what I build

Got a project that needs this kind of thinking?

Tell me what you're trying to build. I'll tell you how I'd approach it.

Start a Conversation