02-347-7730  |  Saeree ERP - Complete ERP Solution for Thai Organizations Contact Us

AI and Accounting

AI and Accounting — What It Can Do Today, and What It Still Cannot
  • 23
  • February

Over the past 2-3 years, the AI (Artificial Intelligence) wave has disrupted every industry, and accounting is among the most talked-about fields. Many accountants worry that AI will replace them, but what is the reality? This article takes an honest look at what AI can actually do in accounting today, and what still requires human expertise.

What AI Can Already Do in Accounting Today

Today's AI technology has several capabilities that genuinely help with accounting work, especially tasks that are repetitive, rule-based, and require high-speed processing of large data volumes. Let's look at each area:

1. Automated Invoice Matching

One of the most time-consuming tasks for accounting teams is three-way matching between Purchase Orders (PO), Invoices, and Goods Receipts (GR) to verify data consistency before approving payments.

AI can help by:

  • Using OCR (Optical Character Recognition) to read data from paper or PDF invoices and convert them into digital data
  • Automatically matching PO, Invoice, and GR by comparing document numbers, quantities, prices, and terms
  • Flagging mismatched items for accountants to review only the exceptions, rather than checking every single entry

The result: tasks that used to take hours each day may be reduced to minutes, dramatically reducing errors from manual visual inspection.

2. Anomaly Detection

AI has the ability to analyze patterns from massive volumes of accounting data and detect anomalies that humans might overlook, such as:

  • Transactions with unusual amounts compared to the average for similar transaction types
  • Accounting entries made at unusual times, such as entries posted at 3 AM when there is normally no activity
  • Duplicate entries that may result from recording errors
  • Transactions that bypass approval workflows or exhibit patterns indicative of fraud

Real-world example:

A company used AI to analyze employee expense claims and discovered that a group of employees consistently filed travel expenses on holidays, with amounts just below the threshold requiring special approval. AI detected this pattern even though each individual claim appeared normal.

3. Cash Flow Forecasting

Cash flow forecasting is critically important for every business, yet often inaccurate due to the many variables involved. AI helps by:

  • Using Machine Learning to analyze historical data such as income-expenses, customer payment behavior, and business seasonality
  • Predicting cash inflows and outflows for each week/month with confidence intervals
  • Providing early warnings when liquidity problems are likely, giving executives time to prepare

The advantage of AI is its ability to update forecasts in real time whenever new data comes in, rather than waiting for monthly forecast cycles.

4. Automated Expense Categorization

When large volumes of expenses come in, categorizing them correctly according to the Chart of Accounts is time-consuming. AI can:

  • Learn from historical patterns — for example, Vendor A was always categorized under "Office Supplies"
  • Suggest categories for accountants to choose from, ranked by confidence level
  • Auto-record high-confidence entries (e.g., 95%+) and route uncertain ones to humans for review

5. Financial Report Summarization

With the capabilities of LLMs (Large Language Models) such as ChatGPT or Claude, AI can now:

  • Read financial statements and summarize them in plain language that non-accounting executives can understand
  • Compare figures between periods and highlight significant changes
  • Answer questions about financial reports, such as "Why did Gross Margin decline?" or "Which expense category increased the most?"
  • Draft preliminary Notes to Financial Statements

Important caveat:

LLMs may "hallucinate" — generating plausible-looking but incorrect information. Therefore, AI-generated report summaries must always be reviewed by an accountant. Never use them directly without verification.

What AI Cannot Do Yet, or Should Not Do

Despite AI's growing capabilities, many accounting tasks still require human professional judgment that AI cannot replicate or should not be entrusted with:

1. Decisions Requiring Professional Judgment

Many accounting tasks do not have a single correct answer and require the judgment of an accountant who understands the business context, such as:

  • Allowance for doubtful accounts — Will this debtor pay? AI only sees numbers, but the accountant knows the debtor is currently in debt restructuring negotiations
  • Estimating useful life of assets — How many years will this machine last? It depends on many factors that AI does not have data on
  • Classifying borderline items — Is this expenditure Capital Expenditure or Operating Expense? It depends on company policy and context

2. Interpreting Accounting Standards

Financial Reporting Standards (TFRS/IFRS) are highly complex and often require interpretation based on each business's specific circumstances, such as:

  • TFRS 15 on revenue recognition — When should revenue be recognized for contracts with complex conditions?
  • TFRS 16 on leases — Does this contract qualify for recognition as a Right-of-Use Asset?
  • TFRS 9 on financial instruments — How should this instrument be classified?

AI may help locate relevant requirements, but interpretation and application still require expert professionals.

3. Complex Tax Planning

Tax planning is not just calculation — it is strategic planning that requires understanding tax laws, business context, and acceptable risk levels, such as:

  • Which business structures offer legitimate tax savings
  • How to maximize BOI investment privileges
  • Transfer Pricing for multinational companies

4. Legal Accountability

No matter how accurate AI becomes, legal accountability still rests with humans:

  • Certified Public Accountants (CPA) must sign off on financial statements — AI cannot sign on their behalf
  • Registered bookkeepers with the Federation of Accounting Professions are responsible for the accuracy of accounts
  • If AI makes accounting errors, liability still falls on humans, not AI

5. Data That Should Not Be Sent to AI

Accounting data is confidential company information. Sending data to AI, especially cloud-based AI, requires careful consideration:

  • Employee salary data — personal data protected under PDPA
  • Undisclosed financial statements — may constitute Insider Information that must not be shared
  • Client and partner data — pricing, terms, contracts
  • Tax data — tax filing details and the company's tax strategy

Critical warning:

Before sending any accounting data to AI, verify that the AI does not use your data for model training and has a clear Data Privacy policy. Otherwise, confidential company data could be leaked.

Summary Table: Which Accounting Tasks Can AI Handle, Assist With, or Not Do?

Accounting Task AI Can Do AI Assists Requires Humans
Invoice Matching
Expense Categorization
Anomaly Detection
Cash Flow Forecasting
Financial Report Summarization
Account Reconciliation
Interpreting Accounting Standards (TFRS/IFRS)
Allowance for Doubtful Accounts (Judgment)
Tax Planning
Signing Off Financial Statements (CPA)
Budget Preparation
Monthly/Annual Financial Closing

Noteworthy AI Tools for Accounting Today

For accountants looking to start using AI, here are some accessible and noteworthy tools:

Tool What It Helps With Pros Cautions
ChatGPT + Excel/Google Sheets Write formulas, analyze data, create templates Easy to use, start immediately Never send actual company data
Microsoft Copilot in Excel Analyze Excel data using natural language Works directly within Excel files Requires Microsoft 365 Copilot License
Google Sheets AI (Gemini) Create tables, formulas, summarize data Free (Google Workspace) Data is stored on Google's Cloud
Claude / ChatGPT (Code Interpreter) Analyze CSV/Excel files, create charts, find patterns Capable of in-depth data analysis Use sample data instead of real data

Getting started tip:

Try starting with ChatGPT to help write complex Excel formulas or explain accounting standards you find hard to understand — without sending actual company data. Use fictitious data instead.

Saeree ERP and AI

Saeree ERP has a comprehensive accounting system covering general ledger, accounts receivable/payable, budgeting, and financial reports that comply with Thai accounting standards.

AI Status in Saeree ERP:

Currently, Saeree ERP does not yet have AI features, but they are in the near-term development roadmap. The development team is researching and designing AI features suited to real-world use by Thai organizations.

What Saeree ERP already has to support future AI integration:

  • Well-structured database — All accounting data resides in a single system with a structure that AI can easily access and analyze
  • Complete Audit Trail — Every change is recorded, enabling AI to learn operational patterns
  • Open API — RESTful API architecture ready to connect with external AI systems
  • Diverse reports — Data is ready for AI analysis from day one of system deployment

When an organization has clean and complete data in an ERP system, applying AI analytics later becomes far easier and more accurate than starting from scratch.

How Should Accountants Prepare?

Rather than worrying about AI replacing them, accountants should prepare themselves to leverage AI as a productivity tool:

  • Learn AI fundamentals — You don't need to code, but you must understand what AI can and cannot do
  • Practice using AI tools — Try using ChatGPT, Copilot, or other tools for everyday tasks
  • Focus on skills AI cannot replicate — Professional Judgment, executive communication, strategic planning
  • Understand data — Accountants who understand and can analyze data will be in greater demand, as they must verify and interpret AI-generated results
  • Uphold ethics — Understand Data Privacy, AI limitations, and professional responsibilities

Conclusion

AI is undoubtedly transforming the world of accounting, but it is not "replacing" accountants — it is "reshaping the role". Repetitive, rule-based tasks will gradually be fully handled by AI, but work requiring Judgment, interpretation, and legal accountability will still need humans.

The modern accountant will no longer be the one who is best at data entry, but the one who uses AI as a tool and turns insights into better business decisions.

AI is not here to replace accountants — but accountants who use AI will replace those who don't. The key is not to fear, but to learn and adapt. Organizations with a solid ERP foundation will be ready for AI far sooner than those still relying on scattered Excel files.

— Saeree ERP Development Team

If your organization needs an ERP system with comprehensive accounting capabilities that is ready for future AI integration, you can schedule a demo or contact our advisory team for further discussion.

Interested in ERP for your organization?

Consult with our expert team at Grand Linux Solution — free of charge

Request Free Demo

Call 02-347-7730 | sale@grandlinux.com

Saeree ERP Team

About the Author

Paitoon Butri

Network & Server Security Specialist, Grand Linux Solution Co., Ltd.