Video summary

เสวนา หัวข้อ "เปิดโลก AI for Accountants: เริ่มต้นอย่างไรให้ทำงานได้จริง"

Main summary

Key takeaways

Educational

Main ideas / lessons conveyed

  • AI will be practical for accountants now, not just a future replacement for jobs.
  • AI is most useful when applied to specific accounting workflows, especially repetitive or time-consuming tasks such as:
    • email handling
    • data extraction
    • report drafting
    • document summarization
  • Adoption mindset matters:
    • Start personally first, then scale to the organization.
    • Make learning “fun/easy” to encourage engagement.
    • Don’t chase tool variety constantly—choose tools and deepen use.
    • Build readiness by preparing inputs/prompts and testing carefully.
  • Successful implementation follows a workflow approach:
    • Map the current process step-by-step.
    • Add AI/RPA/digital tools incrementally.
    • Validate results with checkpoints so accuracy is maintained.
  • AI + RPA + OCR + automation tools can form an end-to-end system:
    • “Eyes” = document reading (OCR)
    • “Brain” = reasoning/writing/summarization (GPT/Gemini/AI chat)
    • “Hands/automation” = executing routine actions (RPA, embedded workflows, Power Automate/Make/automation)
  • Human review remains essential: the content produced by AI should be verified before sending to clients.
  • Security and governance are part of adoption:
    • Be careful with sensitive data (e.g., salary/credit card info).
    • Control how data is used (e.g., paid tiers may allow more control).
    • Avoid training/feeding AI with sensitive or unnecessary “microdata.”
  • Organizational change requires “winning hearts”:
    • Management should try AI first.
    • Use sandbox/free experimentation with low risk.
    • Encourage staff adoption through usefulness, gamification/competition, and positive reinforcement.
  • Training pathways:
    • Mention of a Digital-Age Accounting certificate/course to deepen skills in AI, RPA, OCR, and practical accounting applications.

Methodology / instructions presented

A) How to get started with AI in accounting (mindset + rollout)

  1. Define the real work target
    • Clarify what benefit AI will provide for a specific accounting task (not generic experimentation).
  2. Start with yourself
    • Begin using one AI tool for small, manageable tasks.
    • Learn through usage first; aim for ease and enjoyment.
  3. Focus rather than tool-hopping
    • Choose one main AI (e.g., GPT-type) and go deeper.
    • Avoid frequent switching that causes inconsistency.
  4. Build at the organization level
    • After personal success, expand usage as a whole team.
  5. Prepare and test
    • Invest time to get ready (inputs, templates, prompts, data structures).
    • Use checkpoints to ensure outputs are correct.
  6. Keep humans in the loop
    • AI outputs must be reviewed before being sent to customers.

B) Workflow-based implementation (core automation pattern)

  1. Map the workflow
    • Break the process into steps (e.g., receive customer/data → transform/enter → verify → deliver output).
  2. Identify repetitive/time-consuming parts
    • Often “miscellaneous daily tasks,” such as:
      • email replying/sorting
      • data extraction + formula work
      • drafting routine documents
      • assembling dashboards/reports
  3. Insert tools point-by-point
    • Start with small automation additions rather than rewriting everything at once.
  4. Use AI for “thinking/writing/summarizing”
    • Turn raw data into structured outputs (tables, bullet points, drafts).
  5. Use RPA/automation for “execution”
    • Let software perform clicks/typing, file movements, sending/forwarding, or report generation.
  6. Validate outputs
    • Add verification steps so accuracy doesn’t degrade.
  7. Measure time saved
    • Example outcome described: reduce multi-hour tasks to minutes by combining AI + automation.

C) Practical examples/instructions described for specific tasks

  • Email organization using Microsoft Outlook rules
    • Create rules to:
      • route emails into folders (e.g., by topic: meetings, client-related items)
      • handle actions such as delete/block/accept/sort
    • Goal: reduce inbox clutter and stop reading every email manually.
  • Using GPT-style AI to reduce Excel-heavy work
    • Provide AI with:
      • input files/data (Excel/system extracts)
      • instructions for what to compute/transform
    • Generate structured output, then verify before sending.
  • Build “GPT Projects” (custom knowledge/instructions)
    • Configure a project with:
      • instruction data (standard steps)
      • a template-like workflow
    • Use it to produce customer-specific documents quickly.
  • Automate Google Forms creation using App Scripts
    • Use GPT chat to draft the script logic.
    • Run the script in the Script App.
    • Script exports a ready-to-use Google Form.
    • Goal: reduce manual form creation time.
  • Dashboard generation via automation
    • Use scripts/automation to:
      • pull data from systems
      • produce an updated dashboard view
    • Share via link (avoid sending Excel files).
  • Tax/audit checking concept using AI + domain prompting
    • Guide AI with accounting/audit rules and expected formats.
    • AI checks for inconsistencies (e.g., duplicate entries, formatting/value anomalies).
    • Humans still verify final outputs.

D) AI + automation architecture explanation (conceptual “recipe”)

  • Combine three layers:
    • AI (“brain”): reasoning, summarization, drafting, classification
    • OCR (“eyes”): extract data from PDFs/images/documents
    • RPA/automation (“hands”): operate systems, upload data, send/forward, move files
  • Integration goal
    • A full pipeline that reads documents → interprets/produces text/output → executes required system actions.

Main examples used in the discussion (what AI was applied to)

  • Reducing Excel-based revenue recognition workflows
    • Multi-formula extraction + reconciliation steps using a GPT-assisted approach.
  • Drafting client statements/documents
    • Use AI projects/templates to generate customer documents faster.
  • Creating Google Forms quickly
    • GPT-generated script outputs a form in minutes.
  • Building dashboards for progress tracking
    • Automate progress calculation and sharing.
  • Improving meetings/communication
    • Summarize discussions and prepare structured records for sending.
  • Email payment collection via messaging/link flows
    • Automate reminders and payment status checking (described as a concept/example).
  • General accounting tasks:
    • VAT/purchase tax checks (accuracy, duplicates, anomalies)
    • financial statement drafting and ratio/analysis support
    • identifying “abnormalities” through comparisons (common-size/common-sense style idea)
  • Tool stack examples mentioned
    • GPT, Gemini, OCR concept, Outlook rules, Google Workspace tools, Microsoft/Excel workflows, automation tools (e.g., Power Automate/Make-style “connectors”).

Speakers / sources featured (as mentioned)

  1. Rachit Chairat — host/moderator
  2. Mr. Ariphat Jeeraprasit (referred to as Mr. King / Kungking Arichapat Jirapasit) — CEO of an auditing/accounting firm; speaker on AI + RPA/OCR concepts and accounting applications
  3. Ms. Kamonthip Kedthas (referred to as Ms. Ming / Ming) — leader (organizational growth unit) at MMN Syate Limited / MMN CD; speaker on practical workflow adoption and case studies using GPT/automation
  4. P’Toey — appears as an additional organizer/participant role referenced during the event (question prompting/coordination)
  5. Value Sourcing company — referenced as the organization associated with one speaker’s role/case/context

  6. Professional Accountancy Council / Accounting Council — referenced as organizer/affiliated body (no individual named beyond the above)

  7. ChatGPT / GPT and Gemini — AI platforms referenced as tools (not speakers)

  8. Microsoft Outlook and Excel/Google Workspace — software/tools referenced as features (not speakers)

Original video