The Accountant Who Saw the Future First
How a seasoned finance leader pivoted to artificial intelligence and is now on a mission to bring the entire profession with her.
There is a particular kind of clarity that comes from standing at the intersection of two distinct worlds. Angela Shi, CEO and founder of Empathetic AI, has built her career at the crossroads of two demanding disciplines: the compliance-driven rigor of accounting and finance, and the fast-moving frontier of artificial intelligence. Her Sydney-based startup is now leveraging that dual expertise, transforming how the tax and financial advisory industry operate across the Asia-Pacific region and beyond.
CEO and Founder, Empathetic AI
(Photo: Angela Shi)
Three years ago, when she started telling anyone who would listen that AI was going to transform the tax profession, most people looked at her sideways. Today, her work is reshaping how the tax industry operates. Through Luna, a domain-specific large language model trained on decades of Australian tax legislation and rulings, she has built an AI assistant that helps tax professionals cut through complexity, work faster, and deliver sharper advice to their clients. An idea that once drew sceptical looks has become the direction the entire industry is now moving.
“I was probably the first one in Australia raising that concept,” she says, without a trace of self-congratulation. “Not only tax, but law, all the knowledge workers, all the white-collar work will be transformed. When I first started, nobody agreed. But in only three years, everybody talks about AI now.”
That speed of change is central to everything Angela does, and it is a speed she was uniquely positioned to recognise.
From the Ledger to the Algorithm: A Profile
Angela’s career did not begin in technology. It began, as for so many in this field, with the fundamentals: numbers, compliance, clients, and the particular discipline of mind that accounting demands. Over 15 years, she built her professional life in corporate finance, working across financial reporting, month-end close processes, internal controls, and the exacting rhythm of financial governance. She served in senior finance leadership roles, including APAC CFO at Capital.com and Australian CFO at Plus500, managing teams across seven countries and overseeing nine-figure revenues.
Then came fintech.
Before AI became a mainstream conversation, Angela spent fifteen years at the intersection of financial services and technology, where machine learning had already been quietly at work for years. It was powering credit scoring models, fraud detection systems, and algorithmic trading infrastructure. It was an education in what happens when data-intensive industries stop treating technology as a back-office function and start treating it as a competitive edge.
“When AI became publicly known, it wasn’t a completely new thing to me as a fintech person,” she says. “People who have exposure to technology, especially in fintech and financial services, wouldn’t see AI as a completely new thing.”
But that background also gave her something more valuable than technical fluency. It gave her a translator’s instinct. She understood, from the inside, both the language of accounting and the language of machine intelligence. She could see the gap between them, and she could see, more clearly than most, what it would cost the profession if that gap was not closed deliberately and securely.
In early 2023, she made her pivot. She stepped away from the conventional corporate CFO path and founded Empathetic AI, with a mission that has since evolved into AI advisory, consulting, and finance-grade AI agents building services. She did not wait for the technology to mature, for the professional bodies to issue guidance, or for colleagues to catch up. She started building, learning, talking, and pushing from day one.
“I used to be a finance person in corporate,” she says. “Now I’m in AI, self-taught, working with engineers, teaching people like myself to become AI-native. That’s a new business. I just think there are going to be a lot of new business models emerging in the future.”
Today, Angela serves four distinct customer groups: accounting firms, CFO offices and in-house finance teams, higher education institutions, and accounting and finance students. She collaborates with universities and professional bodies across Australia, runs workshops and advisory engagements, and continues to develop AI tools purpose-built for the profession. She is, in the truest sense, a bridge builder who has lived on both sides of the divide she is working to close.
The Gap Nobody Was Talking About
The company Angela founded began as a technology-led venture, building AI tools purpose-designed for the unique demands of tax and finance professionals. That early product work gave Angela direct visibility into how regulated industries respond to AI, and what it truly takes to make adoption stick.
“The reason for hallucination is that the model captures different, unverified information and makes mistakes,” she explains. “What we do is put guardrails in place and narrow the scope. You can’t just capture information outside of scope. And we have an internal evaluation framework to assess the AI agent’s output.”
It was a smart product. But over time, Angela noticed something that no amount of good engineering could fix on its own.
The tools were ready. The professionals were not.
“The reason people don’t want to adopt AI tools is because they don’t fully understand them, they don’t trust them,” she says. “And because our profession is licence-based, we need to be qualified, we need to sign off the papers, and accountability plays a very important role.”
This realisation gradually shifted the centre of gravity in her business. Building a great AI product for accountants was only half the solution. If accountants lacked the literacy to understand what the tool was doing, to interrogate its outputs, and to apply their own professional judgement on top of it, then the tool itself could not reach its full potential.
The gap she identified is not simply a matter of enthusiasm or resistance to change. It is structural. Accounting is, by its nature, a compliance-oriented profession. A stock trader has every incentive to adopt AI tools that might sharpen their edge because AI directly drives revenue in that context. An accountant working through reconciliations, variance analysis, and tax lodgements is operating in a cost-centre function. The motivation to invest in new tools, and the budget to do so, is often simply not there.
“Some functions are actually supporting functions, some are revenue drivers,” she explains. “If I’m a stock market trader, AI can help me optimise my stock-picking performance. That’s going to directly affect my performance. But in accounting, a lot of functions are cost centres. If it’s a cost centre, the motivation to allocate budget for new tools might not be that high.”
Add to this the nature of compliance work itself, where the instinct is always to reduce risk rather than experiment, and you begin to understand why the accounting profession has moved more cautiously than some other fields.
AI-Plus, or Plus-AI?
One of the most compelling ideas Angela has been exploring is what she calls the blurring of two business models: AI-plus companies, which operate primarily in AI and extend into domains like accounting, and plus-AI companies, which are traditional domain businesses embedding AI into their existing services.
“I’m an AI-plus company. I do AI, and I operate in accounting and finance. You do accounting and you embed AI, so you are a plus-AI company. Those two types of business models will have more and more overlap over time.”
The reason, she argues, is that the knowledge gap between the two is shrinking. As AI literacy becomes more widespread, the accountant who deeply understands both tax law and AI workflows will find themselves in an extraordinarily valuable position, able to advise clients not only on their accounts but on their own digital transformation journey.
This is not merely theoretical for Angela. It is the trajectory her own business has taken. She started by building AI products. She is now providing AI advisory services, consulting engagements, and education workshops to accounting firms, CFO offices, and universities. Today, the service layer around technology has become equally important.
“We help them map out their transformation journey,” she says. “A firm might have twenty people, partners, seniors, juniors, and they want to save 20 or 30 percent of costs and become AI-ready in two or three years. We provide the roadmap, the upskilling, the workshops. And if they’ve already bought a Microsoft Copilot or Claude or OpenAI enterprise licence, we help them use it to its maximum capability, because a lot of people only use about 10 percent of the functions.”
The engagement model is deliberately ongoing, not transactional. In a field where the underlying technology can shift meaningfully every six months, a one-off training workshop has a short shelf life.
“After six months, maybe you have to do a bit of an update. We do a two-hour refresh session. It becomes an ongoing learning journey, until internally you have someone who can be in charge, like a Chief AI Officer, and you no longer need an external party to assist.”
Human in the Loop, For Good Reason
There is a question that shadows every conversation about AI in professional services, and Angela meets it directly. Will there comes a day when clients simply use AI to do their tax returns themselves, bypassing the professional entirely?
“A human is still going to be in charge of the last step, which is the sign-off,” she says. “It’s not only about who does the work. It’s about who signs off the work. AI is not accountable for the mistakes. That’s why we have to be smarter than AI. AI cannot sign off whatever it does.”
This distinction between doing the work and being accountable for it is one that Angela returns to repeatedly. In her view, the rise of AI does not diminish the importance of professional judgement; it actually elevates it. As AI takes on more of the information-gathering, processing, and initial drafting work, the professional’s role becomes more concentrated around the moments that matter most: reviewing, challenging, and ultimately endorsing the output.
“AI is going to create a lot of productivity increase, but also going to make mistakes that humans don’t realize. And we as professionals still need to tidy up the mess at the end.”
The practical implication is significant. Previously, a tax professional might spend hours gathering legislation, reviewing case law, and assembling a preliminary advice document before even beginning to apply their actual expertise. With well-designed AI agents handling that initial stage, the same professional can produce a first-cut draft in minutes and spend their time on the editorial, judgemental work that machines cannot replicate.
“We use AI to accelerate the first stage: information gathering, information summary, information processing. And then on top of that, you apply your judgement.”
The Risk Nobody Has Named Yet
Her years in fintech give Angela a perspective on risk that many in the accounting profession are still developing. For most of her clients’ careers, risk in software has been deterministic. A system either works or it does not. If the trial balance does not balance, there is an identifiable error somewhere in the chain.
Generative AI introduces something qualitatively different: non-deterministic risk. The model does not fail in predictable, auditable ways. It can produce outputs that are plausible, internally consistent, and entirely wrong – without obvious warning – with no error flag, no warning, and no trail to follow. This is why governance, verification, and human review remain central to the way Empathetic AI advises clients and builds agents.
“Previously we all used software and the risks it carries it a deterministic risk. We know if the software doesn’t work, we’re going to have a mistake. The trial balance is not going to balance, easy to identify. But now there is a brand new risk: non-deterministic risk. We’ve never seen this before. How can we govern it?”
This is why Angela has been working closely with universities and professional bodies to develop frameworks and guidelines. The governance infrastructure for AI in professional services is still emerging, and she believes the accounting profession, with its deep culture of audit, internal controls, and external compliance, is well-positioned to lead the conversation, if it can move quickly enough.
“For us as finance and accounting professionals, we do a lot of auditing, internal governance, external governance. Generative AI can raise a whole new category of risks we never thought about before. We’re part of Australian National AI Centre’s Responsible AI Network and I’m very passionate about getting involved in creating frameworks and guidelines and presenting that to professional bodies, educational providers, and business leaders at the roundtable.”
The Profession Reimagined
Angela is not naive about the disruption ahead. She acknowledges freely that junior staff roles are changing, that entry-level tasks are being automated, and that the graduate who joins an accounting firm today will have a fundamentally different career than someone who joined a decade ago.
But she frames this as an opening, not a closing, and she speaks from experience. She is, after all, living proof that a career finance leader can reinvent themselves entirely in the AI era without abandoning the expertise that made them valuable in the first place. The domain knowledge accumulated over years of client work, compliance, and complex financial judgements did not become obsolete when she pivoted. It became her strongest advantage.
Australia is already facing a projected shortage of accounting talent by 2030, partly because younger professionals find the field insufficiently engaging. Angela believes that the infusion of AI into the profession’s daily practice could reverse that perception, transforming it into a genuinely technology-forward career that attracts the kind of technically curious graduates currently drawn to software or data science.
“In Australia, up to 2030, we will have a shortage in the accounting profession. Young people study accounting but don’t want to become accountants because it’s not seen as exciting. But if we can combine the two, accounting and technology, we can make this profession tech-savvy.”
The universities are beginning to move. Professional bodies are beginning to engage. The market, she says, has largely moved past the fear-driven resistance that dominated 2023 and 2024.
“It’s not for each single individual accountant,” she says. “It’s actually for the entire profession. We have to learn, share, and help each other, build the community. Let’s be optimistic. As accounting professionals, we can shape our future together.”
Her long-term vision for her company is, refreshingly, not a fixed point on a horizon.
“I see this as such a dynamic era that everybody needs to be proactive and try to adapt, try to learn. I wouldn’t say the current business model will be 100 percent unchanged in the next five or ten years. I always explore, learn, adapt, and try to find new ways. I’m confident that during this transitional period, I’m more than happy to share my learnings with the profession.”
That, perhaps, is the most useful thing any of us in professional services can take from Angela’s example. Her pivot from corporate finance leader to AI entrepreneur was not a rejection of her past. It was a deepening of it. She brought her domain knowledge with her, applied it to a new set of problems, and turned the translation work itself into a mission.
The accountants who thrive in the next decade will not be those who waited for AI to stabilise before engaging with it. They will be the ones who engaged early, learned openly, and brought their profession with them.
Stay ahead with exclusive insights! Sign up for our mailing list and never miss an article. Be the first to discover inspiring stories, valuable insights and expert tips – straight to your inbox!
