We Built an OpenSource AI Workspace for Finance Because AI Answers Need Evidence

calendar_today ago • schedule3 min read

AI is changing how we interact with data. But in finance, a convincing answer isn't enough.

If an AI assistant tells you that revenue increased by 18%, the next question should be:

Where did that number come from?

Can you trace it to the financial statements, the general ledger, and ultimately the underlying transactions?

That's the problem we're working on with OfinTech Desk, an open-source AI workspace for finance and accounting teams.

GitHub: https://github.com/OfinTech/ofinTech-Dask

The Problem We're Trying to Solve

Finance teams work across accounting systems, Excel spreadsheets, financial reports, and business documents.

Getting an answer often means moving between systems, rebuilding calculations, and manually verifying numbers.

AI makes asking questions easier, but it introduces another challenge: verifying the results.

We believe financial AI should be designed around three principles:

Traceability: Financial figures should be connected to their underlying evidence.

Human control: Important actions should require appropriate approval.

Deployment flexibility: Teams should be able to choose their own AI models and infrastructure.

Introducing OfinTech Desk

OfinTech Desk brings financial analysis, AI assistance, reconciliation, and workflows into one workspace.

  1. Ask Your Finance

Ask questions about your financial data in natural language.

Explore revenue, expenses, profitability, cash flow, and other financial metrics.

The assistant uses financial tools and presents results with supporting information. Users can drill down from financial summaries to underlying ledger records.

  1. Reconciliation and Month-End Close

OfinTech Desk includes workflows for matching bank statement items against the general ledger, identifying exceptions, reviewing variances, and preparing month-end activities.

Rather than automatically posting accounting entries, the system supports human review and approval.

  1. Financial Intelligence

The workspace provides financial statements, KPI analysis, forecasting, cost-centre tracking, and interactive dashboards.

Users can investigate the underlying data instead of relying exclusively on high-level summaries.

Built for Developers, Too

We wanted developers to be able to explore and extend the project without being tied to a single AI provider.

The project supports:

Python-based financial engine and a React-based interface.

A desktop application and an optional Team Server.

Local models through Ollama or LM Studio.

Bring-your-own-key support for multiple AI providers.

Read-only accounting integrations and documented connector interfaces.

Workflow approvals, permissions, and auditability.

OfinTech Desk is available under the Apache 2.0 license.

Try It With Synthetic Financial Data

We included a demo company called Harbor Line Logistics.

It contains fictional accounting records, invoices, budgets, and bank statement data.

The demo includes situations designed to be investigated, including budget pressure, late-paying customers, and reconciliation exceptions.

You can explore it without connecting real business accounting data.

Getting started:

Visit the GitHub repository.

Open the latest release or follow the source installation guide.

Load the Harbor Line Logistics example files.

Explore the executive dashboard and financial statements.

Try tracing figures to their underlying accounting records.

Watch the demo: https://youtu.be/jfSBOAi2N0U

Get started: https://github.com/OfinTech/ofinTech-Dask

What We're Still Improving

Open-source development is an opportunity to learn from engineers with different backgrounds.

We're especially interested in feedback on:

Financial calculation verification and reproducible testing.

Reconciliation accuracy and exception handling.

AI reliability and evidence-linked answers.

Security, architecture, and deployment experience.

User experience for accountants and financial analysts.

The project is actively evolving, and we're interested in finding limitations, improving documentation, and making it easier for contributors to participate.

Join the Project

If you're interested in open-source AI, financial technology, Python, React, or building reliable tools for complex workflows, we'd love your feedback.

You can explore the source code, test the demo, report issues, suggest improvements, or contribute a pull request.

GitHub: https://github.com/OfinTech/ofinTech-Dask

If the project is useful to you, consider starring it to follow its progress.

One question for the developer community:

How would you design an AI system that can explain and verify every financial calculation it produces?

I'd love to hear your thoughts.

🔥 Join developers growing publicly
Share your knowledge, build in public, and grow your developer presence with a global community.

More Posts

How AI Is Changing Web Development, and What Developers Still Need to Understand

Daniel Decker - Nov 10

Your AI Agent Doesn't Need Your Token: Edera's Ariadne Conill on Object Capabilities

Tom Smithverified - Nov 5

Architecting a Local-First Hybrid RAG for Finance

Pocket Portfolio - Feb 25

Dozens of email accounts: learning AI coding tools as a student

rahuldk1105 - Oct 27

Python for commerce students, from a former bio kid

rahuldk1105 - Oct 30
chevron_left
1Posts
0Comments
2Connections
Mohamed Ibrahim Ahmed | Founder & CTO | FinTech & AI Entrepreneur

I am a technology entrepreneur an... Show more

Related Jobs

View all jobs →

Commenters (This Week)

4 comments
3 comments
1 comment

Contribute meaningful comments to climb the leaderboard and earn badges!