How AI is reshaping the accounting industry in 2025
Quick summary:
AI has moved from pilots to the finance tech stack. Accounts payable, expense management, reconciliations, forecasting, and management reporting are the most mature use cases. Adoption is rising but uneven, and new rules such as the EU AI Act introduce concrete obligations in 2025. The winning approach is to pair automation with strong controls and human judgment.
Baltic Assist’s Senior Finance Partner, Kristupas Binkys, shares how our teams use AI today, where it actually saves time, and why an automation mindset matters more than any single tool.
AI in finance function: the 2025 snapshot
"While AI brings many efficiency gains, finance teams should stay flexible and design around the needs of each department rather than forcing one tool everywhere." - Kristupas Binkys, Finance Partner at Baltic Assist
AI has shifted from experimentation to targeted deployment. Finance teams are using machine learning and generative AI to speed up routine work, surface anomalies, and present insights faster, without compromising on control or auditability.
What this means for you
- Faster month‑end close through automated reconciliations and data validation.
- Better visibility with near real‑time dashboards and variance explanations.
- Fewer manual errors and stronger audit trails, if controls are designed up front.
Where AI adds value in accounting today
At Baltic Assist, we combine proven automation with practical AI to streamline core processes.
Document capture and AP automation
- OCR and invoice capture: Corpay and Microsoft Dynamics 365 Business Central with Continia extensions to extract fields, match POs, and code invoices.
- Expense management: Pleo and Kontolink for automatic receipt capture, card reconciliation, and API‑based syncs to ERPs.
Data integration and reporting
- API‑first budgeting and planning: Budget123 for structured driver‑based planning.
- Analytics and dashboards: Power BI plus ERP connectors for interactive reporting, drill‑through, and anomaly flags.
New genAI assistants in mainstream tools
- Microsoft 365 Copilot for Finance: collections support, variance analysis, and guided reconciliations directly in Excel, Outlook, and Teams.
- Xero’s Just Ask Xero (JAX): conversational prompts to raise invoices, query balances, and pull KPIs.
- Sage Copilot: AI features in Sage Intacct that accelerate variance analysis and close support.
- Intuit QuickBooks AI agents: workflow automation for payments, bookkeeping and cash‑flow insights.
"These tools work best when they automate data extraction, streamline expense flows, keep systems in sync, and generate reports that help people decide. If a tool complicates decisions, it is not the right fit." - Kristupas Binkys
Automation beyond the tools
Baltic Assist’s automation team applies RPA and lightweight scripts to remove repetitive steps, reduce hand‑offs, and embed quality checks. The outcome is fewer manual touches and cleaner, faster reporting.
The current situation
Adoption is rising, but uneven
- Large finance teams increasingly use AI for data analysis, reporting, and reconciliations.
- Many small and mid‑sized practices are still at the early stage, often due to budget, skills, or data readiness.
What we see in the field
- Quick wins cluster around AP, expense management, reconciliations, and management reporting.
- The bottlenecks are data quality, change management, and ownership. Clear roles and process maps matter more than the model you pick.
Regulation and risk management you cannot ignore
EU AI Act timelines that affect finance teams
- Entered into force on 1 August 2024 with phased application dates.
- Prohibited AI practices have applied since 2 February 2025.
- General‑purpose AI obligations start applying from 2 August 2025.
- High‑risk system requirements ramp up through 2026–2027.
What this means for accounting and finance
Even if your tools are “off‑the‑shelf,” you are still responsible for governance, documentation, and oversight. Pair AI deployment with:
- A use‑case inventory and simple risk rating for each tool.
- Data protection by design aligned to GDPR, including retention and access controls.
- Human‑in‑the‑loop review for journal entries, reconciliations, and unusual postings.
- Vendor due diligence: model provenance, data handling, uptime and export controls.
- Change control and audit trail: evidence of approvals, prompts used where relevant, and versioning of outputs.
- Model and report validation: sample testing, drift checks, and bias screens where estimates are involved.
Audit and ethics touchpoints to watch
- Auditors and preparers are using genAI to draft, summarise, and analyse, but human oversight and documentation remain mandatory.
- Recent ethics updates emphasise competence, explainability, and safeguards when technology is used in assurance and finance.
Skills and organisational design: adapting to the shift
Multiple studies point to sizable shifts in work. McKinsey estimated in 2017 that up to 14 percent of the global workforce - roughly 375 million people - may need to switch occupational categories by 2030, and a 2023 update projected about 11.8 million workers in the United States alone may need to transition by 2030 as automation and generative AI reshape demand. For finance, the shift is already visible on the ground: repetitive work is shrinking while analysis, storytelling, and control design expand.
Skills to prioritize
- Process and data literacy: map inputs, owners, and systems.
- Prompting for outcomes: write clear, testable prompts and acceptance criteria.
- Analytics and visualization: build explainable metrics and narratives.
- Controls design: segregation of duties, exception handling, and evidence.
Structure that helps
- A small automation chapter that partners with Accounting, FP&A, and IT.
- A product owner for each process area, responsible for backlog and adoption.
A simple AI governance playbook that anyone can follow.
Balancing human analysis with AI outputs
AI is powerful at scanning large data sets, spotting patterns, and generating options. Human judgment is essential for context, materiality, and decisions.
"Reports exist to help people decide. If an AI‑enabled tool makes that harder, use a simpler approach, even if that is a well‑designed Excel template." - Kristupas Binkys
Practical guardrails
- Keep decision ownership with people. Use AI to propose, not to commit.
- Require source links and reasoning for genAI‑produced commentary.
Archive inputs and outputs for auditability.
A pragmatic 90 day roadmap
- Pick 3 use cases with clear ROI. Common starters: AP invoice capture, expense reconciliation, and monthly variance explanations.
- Stand up controls: define reviewers, thresholds, and evidence before go‑live.
- Pilot and measure: baseline cycle time and error rates. Target 20 to 40 percent time savings on manual steps.
- Train the team: short, role-based training and cheat‑sheets.
- Scale in waves: retire old steps as you automate. Keep a backlog of improvements.
Looking ahead
AI is moving from point features to agentic workflows that string tasks together across systems. Expect more proactive alerts, continuous close capabilities, and richer scenario planning. The finance teams that win will be those that combine automation, controls, and storytelling to drive decisions.
Frequently asked questions
What is AI in accounting?
Software that learns from data to automate routine tasks and highlight risks or opportunities. Examples include invoice capture, anomaly detection, and narrative report generation.
Is AI allowed in statutory reporting or audit?
Yes, but with conditions. You must keep human oversight, maintain evidence, and follow relevant professional and regulatory rules. If a tool is high‑risk under the EU AI Act, additional obligations apply to providers and users.
What are the fastest time‑to‑value use cases?
AP capture and coding, expense reconciliation, and automated variance commentary in management reports.
How do we avoid AI mistakes in finance?
Validate outputs, restrict access, log activity, and keep humans in the approval path for postings and disclosures.
Have a question?
Get in touch!
Baltic Assist provides a comprehensive outsourcing solutions that saves costs, enhances efficiency, and strategic decision-making for your business.
