I pasted my bank statements into ChatGPT. Then I built something better.
TL;DR:
I built Ask Linc after literally pasting screenshots of my bank accounts into ChatGPT (🫣). It connects to your real accounts via Plaid, uses GPT-4, and answers questions like “Can I afford to take a break from work?” or “Are my CDs earning enough?”
Here's a peek into how it works and what powers it under the hood.
🧨 The Problem
After getting laid off earlier this year, I was trying to figure out:
How long we could last without my income
Whether our cash was earning enough
If we were still on track for retirement
Naturally, I turned to ChatGPT — but it had zero context about our actual finances. So I did what no one wants to admit:
📄 Uploaded statements
🖼️ Pasted screenshots
🤷♂️ Hoped no one was watching
The answers were decent, but the process felt sketchy and tedious. So I built the tool I wish I had.
🤖 What Ask Linc Does
Ask Linc is like ChatGPT — but it actually knows your financial situation.
You connect your accounts (via Plaid), and Linc gives thoughtful answers to natural-language questions like:
“How long could we cover expenses if one of us lost income?”
“What’s our monthly burn?”
“Should I refinance right now?”
“Are we sitting on too much idle cash?”
It also pulls in live market and economic data to ground its advice in real-world context — not just AI hunches.
🛠️ Tech Stack Overview
Frontend:
Next.js 15 (App Router) + TypeScript + Tailwind, deployed on Vercel
Backend:
Node.js + Express, Prisma + PostgreSQL, hosted on Render
Integrations:
Plaid for account + transaction data
FRED for Fed economic indicators
Alpha Vantage for real-time market and yield data
GPT-4 (OpenAI) as the reasoning engine
Security:
Token-based access, encrypted end-to-end
No sensitive data stored
GDPR compliant + user deletion supported
CI/CD & DX:
GitHub Actions, Vercel + Render auto-deploy, strict TypeScript, Jest + Supertest
🧪 Privacy-First by Design
I'm building this for people like me — privacy-conscious, skeptical of fintech data grabs. So:
No sensitive data is stored
No user data is used for model training
No creepy tracking (we use Plausible, not GA)
Full delete-on-request capabilities
🧠 Why This Stack?
TypeScript end-to-end keeps bugs down
Prisma makes DB ops cleaner + safer
GPT-4 gives powerful, flexible language understanding
Render + Vercel keep infra dead simple
Privacy and performance come first
📈 Where We're At
Live beta with early testers
Sub-second response times
99.9% uptime
Scalable architecture ready to grow
🙋♂️ Looking for Feedback
If you’ve ever tried to ask ChatGPT about your finances — or thought about combining AI + Plaid — I’d love your feedback.
Or just drop a comment and let me know:
What would you want an AI assistant to tell you about your money?
