AI in Treasury: No Code Required
Practical AI workflows for treasury professionals who don't want to become programmers.
Ten copy-paste-ready prompts, the four tools you already have access to, how to write prompts that actually work, when to trust AI and when not to, and a 30-day plan to make it a habit.
If you can use Excel, you can use AI.
- ✓ Use Copilot, ChatGPT, Claude, and Power Automate for real treasury tasks — no coding required
- ✓ Copy, paste, and adapt 10 ready-to-use AI workflows, from cash analysis to automation
- ✓ Write clear prompts using the CRISP framework
- ✓ Know exactly where AI helps and where it can quietly get you into trouble
- ✓ Follow a concrete 30-day plan to build the habit
What You'll Learn Today
Excel Copilot, ChatGPT, Claude, Power Automate
Copy, paste, and start using today
Get better results from AI with simple techniques
Know the limitations to avoid mistakes
Step-by-step implementation plan
AI Tools You Can Use Right Now
No installation, no coding, no IT department needed
Where: Built into Excel, Word, Outlook
Cost: Included with M365 Copilot license
Best for: Data analysis, report writing, email drafting
"Analyze this cash flow data and identify trends"
Where: chat.openai.com
Cost: Free tier available, Plus = $20/mo
Best for: Analysis, explanations, document drafting
"Explain these bank fees and suggest negotiations"
Where: claude.ai
Cost: Free tier available, Pro = $20/mo
Best for: Complex analysis, long documents, reasoning
"Review this treasury policy and identify gaps"
Where: Built into Microsoft 365
Cost: Included with M365
Best for: Automated workflows with AI actions
"Auto-categorize incoming bank statements"
10 Ready-to-Use
AI Workflows
Copy the prompts. Paste into the tool. Get results.
Analyze your cash position data with natural language
Step 1: Export your cash position to Excel — daily balances by entity/currency from your TMS or bank portals.
Step 2: Use this prompt
1. Which entities have excess cash above €500K?
2. Which entities are below €100K and may need funding?
3. What's the trend over the past 30 days - are we accumulating or depleting?
4. Any unusual patterns or outliers I should investigate?
5. Summarize in a table I can share with my CFO."
- • Entity-by-entity analysis with recommendations
- • Trend identification (improving/declining)
- • Flagged anomalies for investigation
- • Ready-to-share executive summary table
Turn bank fee statements into negotiation ammunition
Step 1: Upload your Account Analysis statement (PDF or paste text)
Step 2: Use this prompt
1. List the top 10 fees by cost (show service code, description, volume, unit price, total)
2. Identify fees that increased >10% vs typical rates
3. Flag any duplicate or unusual charges
4. Calculate the effective rate per transaction type (wires, ACH, etc.)
5. Suggest which 3-5 fees I should negotiate first based on potential savings
6. Draft talking points I can use in my next bank review meeting"
Pro tip: Upload 3-6 months of statements to identify trends and seasonal patterns
Typical result: 15-25% fee reduction opportunities identified
Create a CFO-ready FX exposure report in minutes
Input: Your cash position data with currency breakdown
1. Calculate total exposure by currency (converted to EUR at today's rates)
2. Show net long/short position for each currency
3. Highlight currencies where exposure exceeds €1M
4. Calculate what a 5% adverse move would cost us in each currency
5. Create a simple chart showing exposure by currency
6. Recommend which exposures should be hedged based on materiality"
| Currency | Exposure | Position | 5% Risk | Recommendation |
|---|---|---|---|---|
| USD | €2.3M | Long | €115K | Hedge |
| GBP | €1.1M | Short | €55K | Monitor |
| CHF | €0.4M | Long | €20K | Accept |
Build a rolling forecast from historical data
Input: 12+ months of weekly cash flow history
1. Identify seasonal patterns (month-end, quarter-end, holiday periods)
2. Calculate the average weekly cash flow by category (receipts, payroll, AP, etc.)
3. Project the next 13 weeks based on historical patterns
4. Add columns for: Week, Projected Receipts, Projected Disbursements, Net Flow, Cumulative Position
5. Flag any weeks where projected ending cash falls below €2M
6. Include a confidence range (optimistic/pessimistic scenarios)
7. Create a line chart showing the forecast with confidence bands"
⚠ Important: AI-generated forecasts are starting points. Always validate with known upcoming events (payroll dates, large payments, seasonal adjustments) and adjust accordingly.
Enhancement: Ask Copilot to "add a column for manual adjustments and a notes field for each week"
Generate executive-ready reports from raw data
Input: Paste your month-end data (cash positions, transactions, KPIs)
EXECUTIVE SUMMARY (3-4 bullet points of key highlights)
CASH POSITION
- Month-end total cash by region
- Change vs prior month with explanation
- Variance vs budget
LIQUIDITY
- Days cash on hand
- Available credit facilities and utilization
- Short-term investment portfolio summary
KEY ACTIVITIES
- Significant transactions
- FX hedging activity
- Bank relationship updates
OUTLOOK
- Cash forecast for next 30/60/90 days
- Known large cash events
- Risks and mitigation actions
Format it professionally with clear headers and concise language."
Time saved: 2-3 hours per report → 15 minutes of review and polish
Audit your treasury policy against best practices
Input: Upload your current treasury policy document
1. COMPLETENESS CHECK — list standard treasury policy sections and indicate if each is present or missing:
☐ Cash management ☐ Investments ☐ Debt management ☐ FX risk
☐ Interest rate risk ☐ Counterparty risk ☐ Bank relationships ☐ Controls
2. GAPS & RISKS — identify any gaps that could expose the company to risk
3. BEST PRACTICE COMPARISON — compare key limits/thresholds to typical market practice (e.g., investment duration, credit ratings, concentration limits)
4. OUTDATED ELEMENTS — flag any references that may be outdated (old regulations, defunct benchmarks like LIBOR)
5. RECOMMENDATIONS — provide 5 prioritized recommendations to strengthen the policy
Present findings in a format suitable for sharing with internal audit."
Use case: Annual policy review, audit preparation, new hire training
Draft professional emails to banks in seconds
"Draft a professional RFP email to 3 banks requesting proposals for cash management services. Include: company background (€500M revenue, 15 countries), current services we need (pooling, payments, collections), timeline (responses in 3 weeks), and evaluation criteria."
"Write a diplomatic but firm email to our bank relationship manager requesting a fee review meeting. Mention we've identified €50K in annual fees that seem above market rates, we value the relationship, and we'd like to discuss optimization."
"Draft an email requesting an amendment to our revolving credit facility to extend maturity by 2 years. Maintain professional tone, reference our strong payment history, and request a call to discuss terms."
Always review and personalize before sending — AI drafts are starting points!
Find and explain reconciliation differences
Input: Bank statement + GL/TMS transactions in separate columns
1. Match transactions by amount and date (±2 days tolerance)
2. Flag unmatched items on both sides
3. Group unmatched items by likely cause:
- Timing differences (recorded in different periods)
- Amount differences (partial payments, fees)
- Missing on one side (not recorded)
4. For each unmatched item, suggest the most likely explanation
5. Calculate total reconciled vs unreconciled amounts
6. Create a summary table showing reconciliation status"
- • "Find transactions that might be the same item split across multiple entries"
- • "Identify fees or adjustments the bank applied that we didn't record"
- • "Create VLOOKUP formulas I can reuse next month"
Create training materials and process documentation
"Create a step-by-step process document for daily cash positioning. Include: who performs each step, system screenshots placeholders, checkpoints and controls, what to do if something goes wrong, estimated time per step"
"Create a 10-question quiz to test new treasury analysts on: cash pooling concepts, payment processing controls, FX hedging basics. Include multiple choice answers with explanations for wrong answers."
"Explain interest rate swaps to someone who has never worked in treasury. Use simple analogies, avoid jargon, and include a real-world example. Then create 3 practice scenarios they can work through."
Perfect for: Onboarding, cross-training, audit documentation, knowledge transfer
Automate repetitive tasks without coding
Power Automate has AI actions built in — no coding needed
Trigger: Scheduled (6 AM)
Get Excel file → AI: Summarize key figures → Send email to team
Trigger: When file arrives in folder
AI: Extract key data from PDF → Add to Excel log → Flag anomalies
Trigger: When payment >€100K pending
AI: Describe transaction → Check against policy → Send Teams alert
Trigger: Friday at 4 PM
Collect data from 3 sources → AI: Write summary → Create Word doc → Email
To start: Go to flow.microsoft.com → Create → Describe what you want to automate in plain English
How to Write
Better Prompts
Get dramatically better results with these simple techniques
The CRISP Framework for Prompts
"You are a treasury analyst at a €500M manufacturing company..."
"Act as an expert in bank fee analysis..."
"Analyze this data and identify the top 5 cost drivers..."
"Present in a table with columns: Item, Amount, % of Total, Recommendation"
"Write for a CFO audience, keep it under 1 page, professional tone"
Good vs Bad Prompts: Real Examples
"Analyze my cash data"
- • Vague output, may miss what you need
- • Requires follow-up questions
- • Generic advice
"Analyze this cash position data for our 12 European entities. Identify which entities have excess cash above €500K that could be pooled centrally..."
- • Targeted output you can use immediately
- • Complete analysis in first response
- • Specific recommendations for your situation
Weak: "Write an email to the bank"
Strong: "Draft a professional email to our J.P. Morgan RM requesting a meeting to discuss Q1 fee increases. Mention we've been a client for 5 years and want to explore optimization."
Weak: "Help with forecasting"
Strong: "Using this 12-month cash flow history, create a 13-week forecast. Account for month-end patterns and flag weeks where projected cash falls below €2M."
When AI Works (And When It Doesn't)
- • Analyzing and summarizing data you provide
- • Drafting documents, emails, reports
- • Explaining complex concepts simply
- • Finding patterns in historical data
- • Creating formulas and calculations
- • Generating ideas and checklists
- • Formatting and restructuring information
- • Real-time market data (it's not connected)
- • Specific bank rates or fees (may be outdated)
- • Regulatory requirements (laws change)
- • Company-specific policies (it doesn't know them)
- • Future predictions (pattern-matching, not predicting)
- • Confidential data (don't share sensitive info)
- • Mathematical calculations (double-check!)
Golden Rule: AI is a smart assistant, not an oracle. Always review, verify, and apply your professional judgment before acting on AI outputs.
🔒 Data Privacy: Check your company's policy on using AI tools. Avoid sharing confidential customer data, account numbers, or proprietary information with public AI tools.
Your First 30 Days with AI
A practical implementation plan
- ☐ Sign up for ChatGPT or Claude (free tier)
- ☐ Try 3 prompts from this presentation
- ☐ Analyze one real dataset from your work
- ☐ Note what worked and what didn't
- ☐ Use AI to draft one email or document
- ☐ Try Excel Copilot for data analysis
- ☐ Create one forecast or projection
- ☐ Share a win with a colleague
- ☐ Identify your 3 most repetitive tasks
- ☐ Create AI prompts for each
- ☐ Save prompts for reuse (prompt library)
- ☐ Measure time saved
- ☐ Try Power Automate for one simple flow
- ☐ Train a colleague on what you learned
- ☐ Document your top 5 use cases
- ☐ Plan next month's AI projects
Success metric: By Day 30, you should save at least 2 hours per week on routine tasks
Quick Reference: Copy & Paste Prompts
"Summarize this [data type] highlighting the top 3 insights and any anomalies"
"Write a professional email to [recipient] about [topic]. Keep it concise and actionable."
"Explain [concept] to someone with no finance background. Use analogies."
"Review this [document] and identify any errors, inconsistencies, or missing elements."
"Create a checklist for [process] including all steps and common mistakes to avoid."
"Compare [Option A] vs [Option B] for [use case]. Create a pros/cons table."
"Calculate the impact of [change] on [metric]. Show assumptions and sensitivity."
"Reformat this data as an executive summary with key metrics, trends, and recommendations."
Tip: Save these in a document for quick access. Customize brackets [like this] for your specific needs.
Start Small.
Start Today.
→ You don't need to code — just type
→ Start with one workflow that saves you 30 minutes
→ AI makes you more valuable, not replaceable
→ The best prompt is the one you actually use
Your first task: Try Workflow #1 (Cash Position Analysis) this week
Key takeaways
You already have the tools: Excel Copilot, ChatGPT, Claude, and Power Automate. No installation, no IT ticket needed.
10 ready-to-use workflows cover the daily grind: cash analysis, bank fees, FX exposure, forecasting, reporting, policy review, vendor comms, reconciliation, training, automation.
The CRISP framework — Context, Role, Instructions, Specifics, Polish — turns vague prompts into targeted output you can use immediately.
AI is a smart assistant, not an oracle: great at analysis, drafting, and formatting; be careful with real-time data, regulations, confidential data, and math.
30 days, one workflow a week: by Day 30 you should be saving at least 2 hours a week on routine tasks.
If you can use Excel, you can use AI. Start with one workflow that saves you 30 minutes.
AI for Treasury: Build Your Own Tools
Ready to go beyond prompts? The INPUT/PROCESS/OUTPUT model for building your own tools — no-code, prompt+API, custom app.
Next step · CourseAI in Treasury: Separating Real from Marketing
The AI vocabulary that matters, five proven use cases, vendor red flags, and a live Vendor AI Scorecard.
treasuryease.com