Empathetic AI unveils Empathetic AI Labs, a finance-grade AI agents suite for tax and beyond

Empathetic AI has outlined a product and research push around what it calls "finance-grade AI", focused on workflow-aware agents for accounting and tax teams in Australia and other regulated markets.

The Sydney-based company argues that general-purpose chatbots have struggled in professional finance settings because users need verifiable outputs, clear reasoning, and controls that reduce the risk of unreliable responses. It frames its approach as a shift to domain-specific agents that fit into established workflows such as tax research, reconciliation, and compliance review.

From chatbots to agents

Empathetic AI uses the term finance-grade AI to describe systems built for regulated financial work. It highlights three design features: "glass-box reasoning", which explains how the system reached an answer; source traceability, which links outputs to documents such as legislation, rulings, or underlying data; and guardrails, described as controls intended to limit inappropriate responses and align outputs with compliance expectations.

This framing reflects a broader debate in the accounting and tax profession about the practical limits of generative AI. Firms have experimented with conversational systems for drafting and summarising, but many have restricted their use in client work because outputs may not reliably cite sources and professionals remain responsible for the quality of advice and filings.

Empathetic AI points to a trend towards human-in-the-loop validation, where staff review and approve AI-assisted work. It also argues that organisations are moving away from single, general systems and towards modular agents assigned to specific tasks. In this model, a tax research agent might sit alongside tools for document processing, reconciliation, and workflow management.

Luna tax copilot

At the centre of Empathetic AI's product line is New Luna, described as an AI tax copilot. Luna is intended to answer tax research questions with source references and citations, and is pitched for work involving rulings and complex compliance scenarios.

The emphasis on citations targets a key constraint in professional use of generative AI: audit teams and regulators often require a clear record of where an assertion came from. Tax practitioners also need to show the basis of an interpretation, particularly when advice draws on administrative guidance, case law, and detailed legislative provisions.

The company also describes a broader suite under the Empathetic AI Labs banner. These agents are designed to integrate with ERP and CRM systems. Empathetic AI links this to "continuous data flow" across finance operations, with agents handling repetitive tasks and producing workflow-linked outputs for review.

Pressure on practices

Alongside its product positioning, Empathetic AI highlights operational pressures in accounting and tax. It points to talent shortages and burnout in firms, particularly during peak reporting and tax periods, and to continuing regulatory change and growing financial data volumes as drivers of process strain.

These factors have led many firms to look for tools that reduce manual work in preparation and review. In tax, the workload includes interpretation, documentation, correspondence, and evidence collection, along with rising expectations that advisers can defend their position under scrutiny.

Empathetic AI argues that generic AI tools fall short when they cannot show reasoning and evidence. It also describes Australia as an early adopter of specialised AI in regulated environments, linking this to the complexity of the local tax system and the strength of governance and compliance requirements.

Advisory use cases

Beyond research and compliance, Empathetic AI describes a "Smart Advisor AI Agent" for what it characterises as high-judgement advisory work. It cites scenarios such as cross-border expansion, restructuring, and complex tax interpretation. The company says the agent produces advice summaries, step-by-step reasoning, explicit judgement points, and defensibility checklists.

This reflects an emerging use case for AI in professional services, where firms are testing systems not only for speed but also for consistency in documentation. Advisory work often depends on how a recommendation is framed, and how risks and assumptions are recorded. Tools that standardise output could change review practices and reduce rework, but they also raise questions about governance and accountability when material judgements remain with the practitioner.

Roadmap areas

Empathetic AI says its roadmap includes multi-agent orchestration for complex workflows, automated compliance trails for audit readiness, real-time monitoring of regulatory change, and the use of institutional knowledge to provide context for outputs.

The company is also emphasising explainability features, which it presents as necessary for trust and adoption in finance teams. Explainability remains contested in AI because systems can produce convincing narratives that do not reflect underlying model behaviour. In regulated work, firms increasingly look for a combination of citations, structured reasoning, and clear control over what sources the system can use.