Blipko
A WhatsApp chatbot that acts as a full-stack digital accountant, allowing users to input financial information via typing, speaking, or photo sharing.
The Spark ✨
I tried them all—complex finance apps, rigorous Excel sheets, and manual notebooks. They all failed for one reason: friction. Standing in line for a coffee, the last thing I want to do is open an app, wait for it to load, navigate three menus, and log an expense.
I realized the only app I open without thinking is WhatsApp. So, I decided to build my accountant right there.
How it Works
Blipko changes the paradigm from "Data Entry" to "Conversation".
You simply text: "Spent 45 on dinner and 15 for the cab home."
In the background, Blipko uses OpenAI to parse your natural language, identifying multiple transactions, assigning categories (Food, Transport), and recording the time. It replies instantly with a confirmation and a playful emoji.
Technical Architecture 🛠️
- Backend: Node.js serverless functions handling webhooks from Meta.
- Intelligence: GPT-4o mini for cost-effective, high-accuracy intent classification.
- Database: PostgreSQL with strict schemas to ensure financial data integrity.
Recent Polish 💎
I recently tackled some critical user experience issues:
- Reduced Latency: Optimized the webhook handshake and processing pipeline, bringing response times down from ~3s to <800ms.
- Context Awareness: Blipko now remembers recent context, so if you correct a category ("Actually, that was for Work"), it understands which transaction you're referring to.
Roadmap 🗺️
The goal is to make Blipko a complete financial companion:
- Visual Charts: Generating spend analysis charts on the fly and sending them as images.
- Group Splitting: "Split 50 with @Sarah for lunch" will automatically handle the math.
- PDF Reports: One-click export for monthly statements.
Gallery
