Empathetic AI launches "finance-grade AI" agents for tax, audit, and workflow control
Empathetic AI has introduced Empathetic AI Labs - a suite of workflow-aware agents built for regulated finance work. The pitch is simple: finance teams need verifiable answers, clear reasoning, and tighter controls than general chatbots can offer.
From chatbots to agents that fit your workflow
The company describes its approach as "finance-grade AI," with three core design choices focused on auditability and control.
- Glass-box reasoning: step-by-step logic you can review and challenge.
- Source traceability: outputs linked to legislation, rulings, case law, or underlying data. For tax teams, that means direct references to materials like the ATO Legal Database.
- Guardrails: controls to reduce off-topic or non-compliant responses, aligned to internal policies and standards.
This mirrors a broader shift in firms: from one general chatbot to a set of task-specific agents inside existing workflows. Think research, document processing, reconciliation, and compliance review - all with human-in-the-loop validation before anything goes out the door.
Luna: an AI tax copilot with citations
At the centre of the release is New Luna, an AI tax copilot aimed at complex research and interpretation. Luna answers questions with citations and source links, helping practitioners show the basis for advice and satisfy audit needs.
The focus on citations matters. Advisors must evidence positions when guidance, case law, and detailed provisions interact. Having a traceable chain of reasoning shortens review cycles and reduces rework.
Empathetic AI Labs: agents that connect to your systems
Beyond Luna, the Labs suite is built to plug into ERP and CRM platforms. The goal is a continuous data flow: agents handle repetitive tasks, produce workflow-linked outputs, and route items for approval.
For finance leaders, that means fewer manual handoffs, clearer ownership, and artefacts that are ready for audit - not reconstructed at the end of the quarter.
Why firms are paying attention
Accounting and tax teams are under pressure: talent shortages, seasonal crunch, frequent rule changes, and more data than ever. Generic chat tools help with drafting, but they fall short when you need evidence, structure, and controls.
Australia is an early testbed for this kind of specialised AI, given the complexity of the local tax system and strong governance expectations. The learnings here tend to set a useful bar for other regulated markets.
Advisory work: higher judgement, clearer documentation
Empathetic AI also outlined a "Smart Advisor AI Agent" for advisory scenarios like cross-border expansion, restructuring, and nuanced interpretations. Outputs include advice summaries, step-by-step reasoning, explicit judgement calls, and defensibility checklists.
Standardising how advice is framed can speed up review and create consistency. The judgement still sits with the practitioner - the difference is a clearer record of assumptions, risks, and why a position holds.
Roadmap highlights
- Multi-agent orchestration for complex workflows
- Automated compliance trails for audit readiness
- Real-time monitoring of regulatory change
- Use of institutional knowledge to ground outputs
- Expanded explainability features for trust and adoption (NIST AI RMF is a useful reference point)
What finance leaders can do next
- Pick high-friction workflows first: tax research, reconciliation, and compliance review.
- Require glass-box reasoning and citations in every output. No source, no go.
- Set review gates and approval roles. Keep a clear human-in-the-loop record.
- Limit sources the agent can use. Prefer internal repositories and vetted legal databases.
- Integrate with ERP/CRM for data lineage and fewer manual transfers.
- Track impact with concrete metrics: cycle time, defect rate, and rework hours.
- Upskill the team on prompts, controls, and review standards. A curated place to start: AI tools for finance.
Bottom line: agents built for regulated work are moving from pilot to production. The firms that win will combine explainable AI, tight controls, and disciplined review - so every answer stands up to scrutiny.
Your membership also unlocks:
