FinplifAI • Chat

FinplifAI is a RAG-based multiagent chat system designed to assist users with accounting and finance queries in accordance with Spanish state regulations. The platform leverages advanced language models and retrieval-augmented generation to provide accurate, contextually relevant financial guidance.

Architecture Overview

FinplifAI implements a sophisticated architecture with the following key components:

RAG System

The core of FinplifAI is built on Retrieval-Augmented Generation (RAG), which enhances LLM responses by:

• Dynamic retrieval of relevant financial and accounting regulations
• Contextual augmentation of prompts with domain-specific knowledge
• Vector storage of Spanish financial regulatory frameworks
• Real-time document processing capabilities

Multi-Agent System

The platform employs a multi-agent architecture that coordinates specialized components:

• Primary conversational agent (finance/accounting expert)
• Legal information retrieval agent
• Code generation agent for financial calculations
• Document processing agent for analyzing uploaded financial documents

Tech Stack

Frontend and Core

• Next.js App Router

  • Server Components (RSCs) for optimal rendering strategy
  • Server Actions for secure, server-side mutations
  • Advanced routing for complex application flows

• AI SDK

  • Framework-agnostic API for LLM interactions
  • Streaming response handling with structured outputs
  • Provider flexibility with support for xAI (default), OpenAI, Groq, and other providers

• shadcn/ui Components

  • Accessible component primitives from Radix UI
  • Styled with Tailwind CSS for consistent design language
  • Customizable component system optimized for finance applications

AI Infrastructure

• Model Providers

  • Primary: OpenAI (o3-mini, o4-mini)
  • Alternative: xAI (grok-2-1212)
  • Specialized: Groq (deepseek-r1-distill-llama-70b)

• RAG Implementation

  • Custom vector storage for Spanish financial regulations
  • Domain-specific embeddings for accounting terminology
  • Hybrid retrieval system combining semantic and keyword search

Data Persistence

• Vercel Postgres powered by Neon

  • Storage for conversation history
  • User profile and preference management
  • Vector embeddings for rapid retrieval

• Vercel Blob

  • Document storage for uploaded financial files
  • Support for multiple document formats (CSV, PDF, spreadsheets)

Authentication

• NextAuth.js
  • Secure authentication flows
  • Role-based access control for enterprise usage
  • Multi-tenant support with organization boundaries

Key Features

• Bilingual interface with primary focus on Spanish language
• Real-time financial calculations with embedded Python code execution
• Integration with Spanish accounting standards and tax regulations
• Document analysis for financial statements and reports
• Responsive design for desktop and mobile access
• Artifact generation for code, spreadsheets, and documentation

Getting Started

Prerequisites

• Node.js 18.x or later
• Vercel account (for deployments and environment variables)
• Access to required AI model providers

Local Development

1. Clone the repository

   git clone https://github.com/poacosta/finplifai.git
   cd finplifai

2. Install Vercel CLI

   npm i -g vercel

3. Link your local instance with Vercel and GitHub accounts

   vercel link

4. Download your environment variables

   vercel env pull

5. Install dependencies

   pnpm install

6. Start the development server

   pnpm dev

The application should now be running at http://localhost:3000.

Environment Configuration

Create a .env.local file with the following variables:

# Database
DATABASE_URL="postgres://..."

# Authentication
NEXTAUTH_SECRET="your-secret"
NEXTAUTH_URL="http://localhost:3000"

# AI Providers
OPENAI_API_KEY="your-openai-key"
XAI_API_KEY="your-xai-key"
GROQ_API_KEY="your-groq-key"

Deployment

FinplifAI is optimized for deployment on Vercel:

vercel deploy