Video summary

How to Train Your Finance Team on AI in 30 Days (Full Guide)

Main summary

Key takeaways

Business

30-day AI upskilling playbook for finance teams (business-focused summary)

Context / goal

  • CFOs want to upskill finance teams on AI without chaos or security risk.
  • The speaker claims experience training finance leaders across Merciless Banks, KPMG, Rakuten, AWS, and “hundreds” of other companies.
  • The video provides a step-by-step 30-day playbook to move from “no idea where to start” to “confident AI users.”

The 30-day framework (6 steps, ~weekly cadence)

Step 0 (Day 0–1): Choose one AI tool + license everyone

Core directive

  • Pick one AI platform and license the whole finance team immediately (avoid tool sprawl).

Tool selection rules (by ecosystem)

  • Microsoft shop: use Copilot
    • Rationale: improved Copilot value with the “right prompt” and “agents,” plus document/work integration.
    • Security posture: uses the same Microsoft security environment as familiar Microsoft tools (Excel/OneDrive/PowerPoint).
    • Data access examples: Copilot can access email, calendar, OneDrive, Excel, PowerPoint (within the Microsoft tenant).
  • Google Workspace shop: use Gemini (included in license)
    • Rationale: integrated access to Google Drive, calendar, Gmail.
  • Smaller/agnostic team: GPT Business
  • Larger organizations (150+ licenses): Copilot Enterprise or ChatGPT Enterprise
    • Requirement: enterprise plans only, with security standards aligned to SOC 2–level controls (speaker references “SOC to type two level” / professional security).

Operational intent

  • Standardize on one tool so training is repeatable and secure.

Week 1 (Step 1): “Eye opener” 1-hour session (training with real finance output)

Why this matters

  • Licensing alone won’t work: people either learn fast or freeze because they don’t know where to go.
  • The speaker runs a structured “first hour” demonstration.

Training method (finance-first demonstration)

  • Teach that AI isn’t just for “writing emails” by running a live financial analysis using a structured 6-step prompting approach (speaker’s framework).
  • The demo emphasizes:
    • Data integrity & validation (“check the data”)
    • Show what data the model used (“show me the file on which you did the analysis”)
    • Strategy for analysis (ask AI to determine the best analytical approach)
    • Calculation + analysis
    • Commentary drafting (verifiable written commentary as a first draft)
    • Use of Gemini Canvas / ChatGPT canvas to make outputs editable
    • Generate an infographic/dashboard via “create” (speaker shows code generation for a dashboard and the resulting performance evolution view)

Actionable deliverable

  • Participants learn prompting + workflow for producing finance artifacts:
    • validated analysis
    • editable narrative commentary
    • dashboard/infographic style outputs

Step 2 (end of Week 1): Hands-on workshop on real company use cases

Core directive

  • After showing “what’s possible,” force practice using their own tasks/files.

Workshop design

  • Create real exercises tied to internal processes and adapted to the company’s context.
  • Expect troubleshooting—use it as a learning accelerator.
  • This reduces the time barrier (“people don’t have time to invest 1–2 hours on their own”) by structuring practice for them.

Step 3: Build momentum + sharing culture (async channel)

Core directive

  • Create a shared channel (e.g., Slack/Teams) for:
    • best use cases
    • questions
    • peer learning

Operational mechanics

  • Add lightweight “rules” and incentives:
    • when someone shares something useful, reward/recognize them (speaker’s example: “who does in our team’s channel?”)
  • Management principle:
    • don’t let “one-person learning” happen—turn answers into group knowledge.

Outcome

  • The finance team becomes faster than other teams due to a compounding learning loop.

Step 4: Keep learning alive (recurring “Lunch & Learn” + sessions)

Problem addressed

  • Async sharing fades after weeks unless a recurring live element exists.

Recurring format

  • “Lunch and learn” every week or every two weeks:
    • informal first half hour
    • last half hour: a team member shares
  • Share could be:
    • a new tool use case
    • an accounting topic
    • AI workflow implementation
    • replaying content (speaker suggests “play a YouTube video together,” then discuss implementation + share scripts)

Deliverable

  • Ensure ideas convert into implementation steps (who will do what next, and with which data).

Step 5 (end of Month / capstone): AI hackathon to build mini “agents/GPTs”

Purpose

  • A short “mini competition” where the team builds something usable—translated to AI.

Best practice tool approach

  • Build with:
    • Copilot agents (speaker mentions custom GPT)
    • or Gemini equivalents
  • Speaker’s reasoning:
    • templates/agents reduce “blank page” effort when repeatedly producing reports in Excel-like workflows.
    • Agents/GPTs let users change one number and get outputs faster than prompting from scratch.

Three expected outcomes

  1. Teach everyone how to build useful custom AI tools (within 2 hours using a framework).
  2. Produce immediate reusable assets for the finance team.
  3. Identify “high ROI” tools worth scaling (possible expansion beyond the hackathon).

Frameworks / playbooks explicitly referenced

  • “30-day playbook” (overall structured program)
  • Speaker’s “6 steps approach” for financial analysis prompting, including:
    • data integrity/validation
    • reveal data source
    • analysis strategy
    • calculations
    • commentary drafting
    • editable/dashboards via canvas
  • Agent/GPT “template vs blank prompt” concept (template-like agents for repeated reporting workflows)
  • Ongoing knowledge loop
    • license → eye opener → hands-on workshop → async channel → recurring live sessions → hackathon

Metrics / KPIs / targets mentioned

  • No explicit business KPIs (e.g., CAC/LTV, revenue targets, margins, turnaround-time reductions) were stated.
  • The only measurable/timeline elements given are process timing:
    • 30-day plan
    • 1-hour “eye opener”
    • weekly or every-2-weeks lunch & learn
    • ~2-hour hackathon build window

Concrete examples & actionable recommendations

Example outputs to aim for

  • Financial analysis with validated data checks
  • Editable narrative commentary
  • Dashboard/infographic generation via canvas + code generation

Actionable recommendations

  • Standardize on one tool + license everyone (Step 0)
  • Start with a banger demo to prove value beyond email writing (Step 1)
  • Run a hands-on workshop using their own company files (Step 2)
  • Create a persistent Slack/Teams learning hub with incentives (Step 3)
  • Use recurring live sessions (Lunch & Learn) so learning doesn’t decay (Step 4)
  • Host an AI hackathon to build mini agents/custom GPTs/Gemini workflows for faster reporting (Step 5)

Investing/markets emphasis

  • Minimal; no detailed market/investing strategy. Focus remains on internal execution and training ops.

Presenter / sources

  • Presenter: the speaker (not named in the provided subtitles)
  • Referenced companies/tools (platforms/examples): Microsoft (Copilot), Google (Gemini), ChatGPT (GPT Business/Enterprise), Copilot Enterprise/Agents, Excel AI agent mode (as mentioned)

Original video