Team Name:
AdelAIde
Team Members:
AdelAIde
The reality
Key requirements are spread across agencies, and the guidance is written in specialist language, making it hard for people to identify what applies to their specific circumstances. Rules are precise; lives are messy. Bridging the two is hard without help.
Our Vision
What MyGovMate does
From life event to clear tax plan
Tell MyGovMate what’s happening (“I’ve been made redundant”), and it gives you a simple plan: what applies, why, a step-by-step checklist, a when-to-do-it timeline, and the exact ATO links/forms.
Lodge at the right time (no guesswork)
MyGovMate checks when your data is actually ready (e.g., employer has finalised STP, myTax pre-fill has arrived). It then nudges you to lodge after that, so you avoid amendments and stress.
Tell us once, reuse everywhere
When you start a new job, the details you provide (TFN declaration, super choice, HELP flag) are reused across related steps, so you’re not typing the same info again and again.
Catches mistakes before you make them
MyGovMate warns you about common traps before you submit, like redundancy vs ETP coding, PSI for contractors, tax residency issues, main-residence CGT on a home you rented out, or deceased-estate timing.
Built for people
Everything’s in plain English, works great on mobile, is accessible, and every answer shows where the rule comes from. If your case is complex, it hands off smoothly to a real person.
As at 30 June 2024, Australia’s population reached an estimated 27.2 million, with 8.6 million people born overseas (ABS, 2024). With a rapidly growing and diverse population, the demand for accessible, equitable public services is increasing, particularly in essential areas like taxation, benefits, and financial assistance.
In a survey of 1,149 people conducted in South Australia (2016), over half reported they didn’t know what online government services were available, and only around 200 respondents felt confident enough to find information on their own. The same survey showed a perceived lack of online support, especially when users had questions about eligibility, requirements, or how to apply. This shows that many citizens feel overwhelmed by long-winded government content, scattered links, and complex policy documents.
To understand the impact of service complexity, we analyzed operational data from Services Australia for Q3 2024–2025. The data captured millions of calls made across key government service lines such as Centrelink, Child Support, and general agency enquiries. It revealed how calls were handled, whether answered by a service officer, managed through automation, or terminated by frustrated users, and the average wait times.
The results were striking:
- Over 4.8 million calls were made to general Agency support lines in one year, with ~3.7 million needing live help.
- Centrelink (Social Security) also recorded nearly 6 million total calls, with many terminated before being answered.
- The Average Speed of Answer (ASA) for some departments exceeded 23 minutes—particularly for Centrelink and Child Support.
Despite self-service platforms, people still depend heavily on direct human assistance. This shows digital platforms aren’t doing enough. The system’s complexity drives unnecessary demand and delays, creating a frustrating experience for users and burdening already stretched government resources.
This systemic complexity becomes even more apparent in the tax domain. To provide insightful analysis, the Taxation Statistics 2021–22 dataset was implemented to calculate total tax lodging and average hours spent. Individuals who self-prepared their tax returns spent an average of 4.9 hours lodging, while even those using registered agents still dedicated 1.5 hours to the process. These hours are not just numbers; they represent working adults, new migrants, students, and families trying to comply with financial responsibilities in a system that is often difficult to understand.
Faced with this burden, many turn to tax agents. But this reliance doesn’t come cheap: In the 2022–23 financial year, individuals spent an average of nearly $400 just to manage their tax obligations— a figure that’s been steadily increasing year after year. This is particularly significant for low- to middle-income earners, who may not be aware of what is being claimed on their behalf and may struggle to verify or understand deductions and offsets applied in their name. This lack of transparency introduces financial and legal risks—especially for those who rely on agents without receiving proper guidance or explanation.
We built MyGovMate, an all-in-one, web and chat solution that turns confusing, multi-agency rules into simple steps you can act on. It’s your plain-English guide through life’s tax moments, and the everyday ones too.
Type a sentence like, “I’ve been made redundant,” or “Starting a new job next Monday,” or “Renting out my place for six months.” In seconds, MyGovMate translates that life event into a tailored tax plan: what applies to you, why it applies, the step-by-step checklist, a timeline with due dates, and the exact official links and forms. One tap on Add to Calendar and your key deadlines are saved.
Short, friendly prompts catch common pitfalls before you submit—like redundancy vs ETP coding, PSI for contractors, tax residency when you move overseas, main-residence CGT when you rent out your home, or deceased-estate timing.
MyGovMate starts with tax, because most life events are tax events (new job, job loss, starting a business, renting/selling a home, marriage/separation, a death in the family, annual returns, BAS, PAYG instalments). But it’s architected as a life-event layer over myGov, ready to integrate with Medicare, Services Australia, Child Support, and state services. One question in, joined-up answers out.
A clean Gradio front end that accepts plain-English questions and returns concise, source-linked checklists. It includes a built-in reminders drawer and confirms every save, with a large reading area for answers and a compact input box.
A structured mapping that links life-event keywords to short answers, required documents, and official gov.au sources with timestamps. It’s loaded once at startup, version-controlled, and drives deterministic, auditable responses.
A precise understanding layer that uses exact keyword matching and simple patterning to recognize intent and extract dates/times. It composes the final message from the knowledge layer and is ML-ready—future retrieval or small classifiers can plug in under rule guardrails.
A lightweight service that parses explicit reminder commands (saving immediately with confirmation) and asks before saving when dates are only implied. It supports add/list/toggle/delete, keeps items grouped by date, and surfaces clear status back to the UI.
We prioritize the protection of customer information. User text is processed locally within the Space: no external LLMs or third-party APIs receive queries, and no personal data is persisted beyond the session state. Answers cite only official, public .gov.au pages, and behavior is deterministic and auditable. This privacy-by-default approach aligns with government expectations and reduces data-handling risk.
The modular design is domain-extensible: beyond tax, the same pattern can incorporate Medicare, Services Australia programs, and other agency workflows (e.g., ABRS/ABR, Home Affairs) with new intents and checklists. If needed, optional ML (BM25/embeddings, small classifiers) can be introduced to improve recall while rules maintain compliance. Runs efficiently on basic CPU hardware across Spaces, Colab, or containers.
Users type natural language and receive concise, friendly steps with links to official sources. When a date is detected, the bot offers a one-click reminder and confirms the save. The interface provides a large reading area and a simple input box, minimizing cognitive load. Consistent templates make multi-event conversations predictable and quick.
Responses are grounded in government publications, shown with titles and update dates. Rule-based composition prevents hallucinations and keeps tone neutral. The knowledge file is versioned and reviewable, enabling transparent governance. Future ML components will operate under these guardrails to preserve factual accuracy.
Description of Use This was one of multiple datasets we gathered from ATO, which we consolidated into a structured format to analyse operational and financial patterns. The data was filtered by financial years from 2017–18 to 2022–23, focusing on the total number of individuals claiming tax management expenses and their corresponding average spending. Unnecessary fields were removed, and values were standardised to ensure consistency across years. This consolidated dataset was then visualised to highlight trends in individual expenditure on managing tax affairs, providing insights into the rising costs faced by taxpayers.
Description of Use This was one of multiple datasets we gathered from Services Australia, which we consolidated into a structured format to analyse telephony service delivery across key categories, including Agency, Centrelink (Social Security), Child Support, and Health-related services. The data was filtered to show the number of calls answered by service officers, answered through self-managed channels, and terminated by customers, alongside the average speed of answer (ASA) in minutes and seconds. Unnecessary or duplicate information was removed, and the dataset was standardised to allow comparisons across service areas. This consolidated data was then visualised to highlight both call volumes and service efficiency, providing insights into customer demand, telephony performance, and operational challenges.
Description of Use This dataset was gathered from ATO and was one of several taxation-related datasets we gathered and compared the compliance burden of tax declaration between individuals who self-prepare and those who use agents. The data was filtered to calculate the weighted average hours spent by each group, revealing that self-preparers spend significantly more time (4.9 hours) compared to agents (1.5 hours). Redundant or outdated details were removed, and the data was standardised for consistency. The final dataset was visualised to demonstrate efficiency differences, offering insights into taxpayer practices,agent-supported lodgments, and areas where digital self-service options could be enhanced to ease the time burden on individuals.
Description of Use This dataset contains the results of a citizen survey conducted in South Australia in 2016, focusing on the digital landscape and the use of online government services. The data captures the views of South Australian citizens on their use of digital government services. It contains multiple survey responses that how they do not feel like they would be able to find the information they need via Government websites. We filtered and structured the data to highlight relevant insights across demographics such as age, income, and employment status, while removing redundant fields to make it analysis-ready. This refined dataset is being used to inform the development of digital service device that improve accessibility, trust, and user experience for South Australian citizens.
Description of Use We use the Australian Government AI Technical Standard as GovMate’s governance backbone: structure docs/processes around the AI lifecycle (Design → Deploy → Monitor), keep a reference architecture and auditable change log, and ensure explainable answers with gov.au sources. When we add ML, we apply the standard’s risk/bias/assurance controls while retaining the current rule-based NLU as guardrails.
Go to Challenge | 12 teams have entered this challenge.
Go to Challenge | 34 teams have entered this challenge.
