Hi Reader,

CFOs keep asking me:
“How do we skill the team on AI without chaos or risks?”

My answer:

• Pick one secure tool fitting your current environment
• Teach real use cases you can repeat
• Create an environment and platform where people learn from each other

Why this works:

We avoid “shadow IT” by selecting one approved AI inside their existing security perimeter (mostly Microsoft 365 or Google Workspace).

Then we run a 30-day enablement: adapted use cases, hands-on prompts, custom finance agents, and I help them build an environment for sharing wins.

This structure consistently drives adoption.

​
Today I am sharing the exact playbook I use with 100s of finance teams when I upskill them on AI:

Step 1 — Week 0: Choose one tool and license everyone

Pick one AI platform that already lives in your security perimeter and give every Finance team member a license:

• Microsoft shop → M365 Copilot (Business/Enterprise)
• Google shop → Gemini for Workspace
• Small team or scattered stack → ChatGPT Teams/Enterprise

This single decision prevents shadow IT, keeps data in existing controls, and connects AI directly to your mail, files, calendar, and SharePoint/Drive.

⚠️ Important: in enterprise-grade plans (Copilot Business/Enterprise, Gemini, ChatGPT Teams/Enterprise), providers do not train on your company data and is covered by at least a SOC 2 Type II level of security. That’s why I insist on enterprise licensing, not Plus/Pro.

Step 2 — Weeks 1–2: Run a 1-hour “Eye-Opener”

This kickoff is short, high-signal, and designed for busy finance people to get hands-on the same day.

This is where you need to break the belief that AI doesn’t work for finance by showing real use cases.

Content outline:

• Live demos of use cases for Finance (FP&A, Accounting, Treasury)
• CSI–FBI prompting framework (Context, Specifics, Instructions, Format, Blueprint, Identity)
• Risks & limitations—what AI can and can’t do, and how enterprise plans mitigate risk
• Drafting long-form docs (policies, tax memos, executive briefs)
• Financial analysis workflows to get real analysis and insights from AI, not basic ones
• AI working for you on Excel for formulas, checks, and charts

When I give these webinars, I also explain how LLMs work and the difference between all of them (Copilot, ChatGPT, Gemini).

​

Step 3 — Week 3: Hands-on Workshop with your use cases

Next, we run a 2-hour, camera-on workshop with exercises, Q&A, and breakout rooms.

Typical modules:

• Prompting for Finance (CSI–FBI + optimization)
• 5-Step Workflow for Financial analysis with reasoning models on real data from your company (using descriptive, diagnostic, predictive, prescriptive analysis with AI)
• Excel Models (AI builds KPI tabs, linked charts & formulas—not hard-coded numbers)
• Automations (Custom GPTs, Copilot Agents, Gems in Geminj)
• Breakout rooms to develop specific use cases
• Automating reporting with AI

Participants leave with:

• Companion booklet
• Prompting Guide
• Full slide deck with extended examples

To give you an example, here is my five-step workflow I teach for Financial Analysis

1. Verify data
2. Propose analyses
3. Calculate
4. Build visuals
5. Write commentary

That’s how AI stops being a chatbot and becomes a reliable junior analyst.

Step 4 — Week 3: Launch a Slack/Teams channel

On the same week, I help Finance teams open a dedicated #finance-ai-wins channel. If you want to do it yourself, the best is to set it up this way:

Pin three posts:

1. House rules (share sanitized snippets only; focus on AI topics only; small wins better than nothing)
2. Use-case template (Problem → Prompt → Output → Impact → Files/Links)
3. Champion roster (two power users per sub-team, reply <24h)

This prevents knowledge from going dark, keeps wins visible, and makes peer learning the default.

​

Step 5 — Every 2 weeks: Lunch & Learn

Lunch & Learn is a one-hour virtual or in-person meeting focused on exchanging, with almost no preparation needed, as only one person needs to present a use case.

Format: 20–30 min demo + 20–30 min Q&A. Make sure to rotate presenters every time.

Example sessions:

• Automating data cleaning or the consolidation of 365 CSVs into one Excel file → showing how AI can help you set up Power Query, VBA, Python, Google or Office Scripts
• Spreadsheet → story: auto-generate executive one-pagers instead of formatting slides (using Canvas function in ChatGPT or Gemini)
• Variance commentaries: turn P&L tables into crisp 200-word CEO updates (drivers, quantification, next actions)

During this lunch and learn this is also the opportunity to discuss wins and measure their impact on capacity & cycle time, not just “hours saved.” Track:

• Time-to-first-insight
• Variance root-cause time
• Forecast turnaround
• Manual steps automated

The first win I see with AI: most teams reinvest saved time into deferred work—so productivity shows up as throughput and quality.

Step 6 — ~1 month after kickoff: Hackathon

Now that the team is fluent, run a 2-hour hackathon to build agents.

Make it easy:

Ask your team to focus on these defining these three elements.

Name → Instructions → Knowledge (2–3 knowledge document or previous examples of what you want to automate).

Winning agents include:

• Monthly Commentary Bot (reads P&L/variance, outputs house-style summary + 3 actions)
• Expense-Policy Q&A Bot (answers approvals/forms from internal policy docs)
• Presentation Builder (drafts decks in brand voice from notes)

Agents can be demoed in Copilot, Gemini, or ChatGPT Enterprise—and shared org-wide.

​

Why this playbook works

After training 100s of Finance teams at Mercedes-Benz, KPMG, Rakuten, AWS and also at SMEs all over the world, this is why this playbook has results:

• Eliminates risk of shadow IT (one platform, licensed for everyone)
• Replaces “upload & pray” with structured workflows that show results bringing new insights
• Creates a public pipeline of wins (channel + Lunch & Learns) so adoption compounds
• Ends with ownership: your people build the first agents they’ll actually use

Case studies:

• At Rakuten (Barcelona), moving from “best prompts” to five-step workflows + reasoning models produced an immediate quality jump
• At Ferrero Headquarters, a structured workshop turned pain-point mapping into solution plans and short management pitches (in a single afternoon)
• At C&A, we launched a program for Finance to fully use Copilot and become the best in class in their group
  ​

How can I help your team

You can run this playbook yourself, or if you need help, you can work with me.

Here’s what I deliver when you bring me and my team of AI Finance Experts in:

• 1-Hour Eye-Opener (company-wide kickoff, tailored to your stack)
• 2-Hour Hands-on Workshop (Finance use cases, exercises, take-home materials)
• Slack/Teams setup (templates + champion charter, plus Lunch & Learn runbook)
• Agent Hackathon (Custom GPTs, Copilot Agents, Gemini Gems - starter kits included)

👉 Reply with “AI Training” and your preferred dates. I’ll send the short agenda we’ll use to upskill your team.

If you don‘t have a big team, the best way is to join the AI Finance Club and learn first by yourself and then help your team by using the playbook above yourself!

My calendar for AI workshops is filling fast. If you want your team AI-ready before year-end, now is the time to contact me.

​
Your AI Finance Coach,
Nicolas
​
​
​P.S. AI training done right is not only about saving hours. It’s about building a team that works smarter, faster, and with more confidence… and which is recognized to be the best in class in your company!

Reply with "AI Training" and your details to make sure I can still help you before the end of 2025 as there are only a few available spots left in 2025.