AI Business Card Lead Conversion System

Live

Visit Live Site Start Similar Project

Timeline

Not documented in repository

Industry

B2B sales / marketing / events / networking

Role

Full-stack product and platform engineering (frontend + backend)

Services

Product UX: dashboard, contacts, automation canvas, campaigns, messaging, analytics
Frontend: Next.js App Router, PWA-related work per repo docs
Backend: NestJS modules (OCR, contacts, campaigns, messaging, AI, scoring, compliance)
Integrations: OCR providers, OpenAI, WhatsApp/email, OAuth, webhooks, optional Clearbit
DevOps: Vercel serverless, proxy/CORS patterns, cron, observability hooks

01 — Challenge

Manual transcription and follow-up from business cards created slow capture, inconsistent contact hygiene, and easy-to-miss or generic outreach after events and networking.

02 — Solution

Built an end-to-end lead conversion platform: card ingest with Google Vision / AWS Textract and optional LLM structuring; unified contacts with deduplication and enrichment; AI-assisted messaging; automation flow builder; lead scoring and analytics; JWT/RBAC and compliance/audit hooks; realtime updates (e.g. Pusher). Frontend and backend run as separate Vercel projects with a same-origin API proxy pattern for browser traffic.

Key Outcomes

01
Delivered a working internal product foundation across OCR, contacts, messaging, automation, and analytics modules.
02
Internal roadmap docs describe strong progress against a broad feature set, with enterprise integrations (CRM, SSO, Zapier) still outstanding.
03
Marketing/CV-style documents list performance figures (accuracy, throughput, engagement, etc.) as narrative claims—they are not verified as live product metrics in the repository.

Project Gallery

Analytics — KPIs, conversion funnel, and email vs WhatsApp channel comparison with export.

Automations — node canvas with triggers, actions, logic, and AI-assisted flow generation.

Implementation Highlights

1
OCR pipeline with cloud providers and LLM enhancement for structured card data.
2
AI layer for intent, entities, and personalization; migration notes toward newer OpenAI model defaults.
3
Multi-channel messaging with templates, sequences, and visual automation canvas.
4
Lead scoring and analytics backed by MongoDB aggregations and dashboards.
5
Duplicate detection, enrichment, and unified contacts (multiple implementation passes documented).
6
Production-focused work: CORS/proxy between origins, async OCR patterns, cold-start and performance documentation.

Tech Stack

Next.js 16
React 19
TypeScript
Tailwind · shadcn
TanStack Query · Zustand
NestJS 11
Node 22
MongoDB Atlas
Vercel Blob
OpenAI
Google Vision · AWS Textract
Twilio WhatsApp
Pusher

Performance Metrics

Docs reference example deployments: frontend on Vercel; backend API deployed separately (verify current ownership).
Quantitative claims in CV-style docs are narrative targets—not measured outputs in application code.
Feature roadmap documentation cites substantial progress with notable gaps (e.g. CRM connectors, SSO, Zapier).

Perfect

Let's talk

30 min intro call

Available for projects

Planning a similar product?

We help teams move from idea to launch with clear strategy, premium execution, and measurable outcomes.

info@adranco.com