// case_study --id=b2bleadfinder
B2BLeadFinder.
A production-grade Google Maps lead-generation tool — it finds local businesses with weak digital presence, pulls their contact details, and drafts AI-powered outreach. Built pre-funding, AI-assisted, and live at b2bleadfinder.io. The exact approach we ship for clients.
// the_brief
What B2BLeadFinder is
B2BLeadFinder scans Google Maps to find local businesses with a weak digital presence — no website, no phone, a thin listing — surfaces their contact details, and drafts AI-powered outreach so an agency or freelancer can pitch them. The whole idea is to point sellers at businesses that visibly need help and are not already buried under everyone else’s cold email. It is our own product, built to validate a sharper take on a crowded category.
We built it the same way we build for clients: pre-funding, on a real budget, AI-assisted, and to a production standard from the first commit. It is live at b2bleadfinder.io and serving real customers — not a demo we spun up to have something to point at.
// our_role
- Product & architecture
- Full MVP build
- AI-assisted engineering
- Deploy & operate
// the_starting_state
The gap in the market
The big lead tools are built for the opposite of this. They index companies with a strong digital footprint — the ones with a website, a LinkedIn, a tech stack to scrape — and they price for enterprise sales orgs. The businesses an agency can most easily help are the ones those tools cannot even see: the plumber, the clinic, the local shop with a Google Maps pin and no website. Finding them, and getting a real phone number or owner to contact, is genuinely hard at any scale.
The opportunity was a focused, well-priced tool that does the core job cleanly. The risk — the one that kills most attempts in this space — was building it cheaply enough to validate without shipping the kind of fragile, data-thin prototype that no real buyer would trust. That tension is exactly what the AI-assisted-with-real-engineering approach is built to resolve.
We also chose it on purpose as the proof point. It would have been easier to showcase a polished marketing site or a simple content app — the kind of thing that looks good in a portfolio and proves little. A lead tool is the opposite: it is data-heavy, it has real accuracy stakes, and a thin version is immediately obvious to anyone who uses it for ten minutes. If the AI-assisted approach could produce something trustworthy here, on a pre-funding budget, the argument for it on a client’s MVP makes itself.
// what_we_built
What we built
A full production application: authenticated accounts, a map-driven search where you set a business type, location, and radius and scan an area; gap filters for “no website” and “no phone”; contact enrichment on the results; AI-drafted outreach; and subscription billing. Real infrastructure, real database, real auth — the standard the rest of the Build Startup offer promises.
The harder half was never the screens — it was the data pipeline underneath. The tool lives or dies on whether it can scan Google Maps for a query, detect which businesses are actually missing a website or phone, clean the contact data, and return it fast enough that the map feels instant. That is the messy, edge-case-heavy part where most thin clones fall apart, and it is where the engineering went. The parts a user sees are the easy half; the parts that make the results trustworthy are the work.
The stack is the same one we default to for client MVPs, precisely because B2BLeadFinder proves it holds up: fast to build, comfortable at the scale a growing tool needs, and easy for any engineer to extend. Nothing here is a one-off we would not repeat for a client — that is the point of building our own product on the exact process we sell.
framework Next.js (App Router) + TypeScript
database Postgres
auth session-based accounts
data Google Maps scan + gap detection
+ contact enrichment pipeline
ai AI-drafted outreach proposals
payments subscription billing
hosting Vercel + CI/CD
built_with Claude Code · Cursor · v0
(engineer-reviewed throughout)// the_ai_assisted_process
How it was actually built
This is the section that ties the case study to the offer. B2BLeadFinder was built with the same loop we describe on the AI-assisted build page — and it is the clearest proof that the loop works in production, not just in a pitch.
Claude Code handled the codebase-wide work: scaffolding the Maps scan-and-filter flow, wiring the contact enrichment pipeline, building out the account system and billing across files. Cursor drove the tight in-editor loops. v0 generated first-pass UI we rebuilt to spec. And at every meaningful fork — the data model, the auth boundaries, how the AI proposal step was prompted and guarded, the parts that would break under real use — an engineer made the call, rejected what did not hold, and added the guardrails the tools skip.
The result is the point: a tool genuinely built at AI speed that does not read as AI-built to the person using it. No fragility, no data-thin shortcuts — production code, shipped fast.
// timeline
Concept to live
First concept in March, live in April — roughly six weeks. And not a thin starter MVP: a live Google Maps scan-and-enrich pipeline, AI outreach, billing, and 60+ SEO pages, shipped at MVP speed. It has been compounding ever since.
Concept
Mar 2026
The sharper take on the category, scoped to a buildable v1.
Build
Mar–Apr 2026
AI-assisted engineering — the scan pipeline, app, auth, AI proposals, and billing.
Launch
Apr 2026
Deployed to production at b2bleadfinder.io.
Live & growing
Since Apr 2026
Serving real customers; refined on real usage and compounding in search.
// outcomes
The outcome
Total clicks
3.56K
Total impressions
85.2K
Avg CTR
4.2%
Avg position
9.3
First 28 days of organic search — zero paid acquisition.
The win is not just that it shipped — it’s that it compounds. A real product in market, built on a pre-funding budget, now serving 2,345+ users across 80+ countries, grown entirely through organic search with zero paid acquisition. That is the difference between a demo and something you can run a business on.
“We built B2BLeadFinder because every agency struggles with the same problem — finding businesses that actually need their services. Instead of cold-calling random contacts, our tool scans Google Maps and shows you exactly who needs help.”
// what_it_proves
This is what we ship for clients too
B2BLeadFinder is not a portfolio piece we polished for show. It is the approach, run on ourselves: a pre-funding budget, AI-assisted speed, real engineering, a production-grade result that’s live, serving customers, and growing in search. When we tell a founder we can build something real — fast — that survives real users and keeps compounding after launch, this is the evidence, inspectable at a live URL, not a claim on a slide.
It also means our incentives are honest. We carry the same costs we quote you, we made the same scope trade-offs under the same budget pressure, and we live with the result every day in production. An agency that has never shipped and operated its own product on this model is guessing; we are not. That is the difference we want you weighing when you decide who builds the thing your company depends on.
// build_yours
Want this for your idea?
We built B2BLeadFinder pre-funding, the Build Startup way. We can build yours the same way. Start with a scoping call and we’ll tell you honestly what your MVP looks like as a real product.