Team Name:
NeuroGov
Team Members:
NeuroGov
TrueNorth NT (Financial AI Chatbot)— Your Trustworthy Data ChatAgent
Overview
Our project is an AI-powered financial chatbot built to make government budget data simple and accessible. Using the Australian Government Budget 2024-25 and Portfolio Budget Statements (PBS) dataset, the chatbot allows Northern Territory Government (NTG) finance staff to query financial statements in plain English.
Instead of searching through spreadsheets, employees can ask questions like:
– “What is the revenue for 2024-25?”
– “Show me the cash flow for 2025-26”
– “Compare expenses across 2024-25 and 2025-26.”
The chatbot then delivers accurate, transparent answers directly from official data, saving staff time and improving decision-making.
How It Works
• Normalised Data: Budget spreadsheets were refined with column and row mappings to recognise fiscal year formats (2024-25, FY2024-25) and synonyms (revenue, receipts, income).
• NLP Layer: During training, the chatbot learns to classify user questions, determine the correct metric and year, and generate SQL queries. Synonym lists and mappings were refined iteratively to handle diverse phrasing and corner cases.
• Chat Interface: Users access the system through a web app with a responsive chat UI, where they can type questions and see instant answers with confidence scores and suggested follow-ups.
• Backend System: Built on Django and Django REST Framework, with a role-based user model (admin/user), logging, and an audit trail for accountability.
Benefits
• Efficiency: Staff no longer need to manually search large PBS spreadsheets.
• Accuracy: Answers are grounded in official budget documents.
• Accessibility: Even non-technical users can retrieve complex financial data.
• Scalability: Future budgets and additional datasets can be integrated.
Data Security & Sovereignty
A key feature of the chatbot is that it operates fully within the NT Government ecosystem. Unlike consumer AI tools that transmit information to external servers, this system ensures that sensitive financial data never leaves government control.
• No external AI services are used; all processing occurs within secure, NTG-managed infrastructure.
• Data sovereignty is preserved, giving NTG full ownership and control of its financial datasets.
• The system is compliant with public sector requirements around confidentiality, governance, and risk management.
By keeping the data entirely inside government systems, NTG can confidently use AI to improve workflows without compromising security or privacy.
Future Growth
While the prototype is built on the Budget 2024-25 and PBS dataset, the chatbot is designed to be scalable across other NT Government agencies. By applying the same data-normalisation and NLP mapping approach, it can be extended to:
• Freedom of Information (FOI) statistics – enabling quick insights into transparency and compliance data.
• HR datasets (workforce, leave tracking, benchmarking) – supporting HR teams with instant answers on staffing, workforce planning, and employee wellbeing.
• Other tabular benchmarks and public datasets – unlocking faster reporting and data-driven decision-making across departments.
• Voice Integration – enabling staff to interact with the chatbot through speech, making it even more accessible and user-friendly across devices.
• NTG SharePoint Integration – embedding the chatbot directly within departmental SharePoint sites, allowing staff to query datasets inside their existing workflow tools without switching platforms.
This positions TrueNorth NT as a versatile AI assistant for the NT Government, capable of growing beyond finance to streamline information access in multiple areas.
Vision
TrueNorth NT lays the foundation for smarter government data tools. With future enhancements like data visualisation, cross-year comparisons, SharePoint integration, voice access, and reporting features, this system can evolve into a comprehensive assistant for public sector teams across finance, HR, and beyond.
Description of Use We have selected only one here to show how the model is going to work.The AI Budget Chatbot is designed to make the Australian Government Budget 2024-25 and Portfolio Budget Statements (PBS) dataset instantly accessible for NTG finance staff. The dataset contains large, complex spreadsheets covering revenue, expenses, assets, liabilities, and equity across multiple years (2024-25, 2025-26, 2026-27). Traditionally, staff must manually locate and interpret figures from these tables, which is time-consuming and prone to error. With our model, each table was normalised and enhanced with: Column mapping: recognises fiscal year formats (e.g. 2024-25, 2024/25, FY2024-25) and metric synonyms (e.g. revenue, receipts, income). Row mapping: links common user queries (e.g. “What are total expenses for 2024-25?”) to the correct table rows. During training, the NLP layer uses these mappings to classify a user’s question, determine the metric and year, and generate the corresponding SQL query. Refinement involved iteratively expanding synonym lists, testing corner cases, and ensuring every important row and column could be matched from natural language. In practice, NTG employees will use the chatbot to: Ask plain-English questions such as “Show me the cash flow for 2025-26” or “What are the equity changes between 2024-25 and 2025-26?”. Retrieve accurate results directly from the PBS dataset, presented in a simple response format. Compare multiple years or metrics without manually building queries in Excel. By grounding all answers in the official Budget 2024-25 dataset, the chatbot provides reliable, transparent, and context-aware insights. This use case allows finance staff to spend less time navigating spreadsheets and more time on analysis, planning, and decision-making.
Ask your data, trust the answer
Eligibility: Open to all, but special consideration will exist for teams with a local NT lead.
Go to Challenge | 24 teams have entered this challenge.
