Every finance leader knows the feeling: you sit down with a fresh set of monthly statements, and the cash conversion cycle has stretched by three days. Accounts receivable is bloated with overdue invoices. Inventory is piling up in slow-moving SKUs. Payables are being stretched to the limit, and your suppliers are starting to complain. You run the same old Excel models, pull the same reports from your ERP, and yet the working capital metrics barely budge. The friction is real—manual data extraction, siloed information, and a lack of real-time visibility into the drivers of cash flow. You know the levers exist, but pulling them in the right sequence feels like guesswork.
This is where ChatGPT changes the game. Instead of spending hours reconciling spreadsheets or building static dashboards, you can now use a conversational AI to analyze your financial data, identify patterns, and generate actionable recommendations in minutes. ChatGPT can ingest your AR aging reports, inventory turnover data, and payment terms, then produce a targeted optimization plan. It doesn’t replace your judgment—it amplifies it. By asking the right questions and structuring your data, you can turn a messy set of numbers into a clear, prioritized roadmap for freeing up cash.
In this post, I’ll show you exactly how to use ChatGPT for working capital optimization. You’ll get two ready-to-use prompts that follow a structured “anatomy” approach. These prompts are designed to be copy-pasted into your ChatGPT session with your specific data. The first prompt focuses on diagnosing your current working capital position. The second prompt generates a tactical improvement plan. Let’s dive in.
Why Most Working Capital Analysis Fails
The root cause of poor working capital optimization isn’t a lack of data—it’s a lack of structure. Finance teams typically pull reports from different systems: AR from the billing module, inventory from the warehouse management system, and AP from the procurement platform. These datasets rarely speak to each other. Even when they do, the analysis is often backward-looking, showing you what happened last month rather than what you should do next week. ChatGPT solves this by acting as a unifying layer. You feed it the raw data, and it applies financial logic to surface the biggest opportunities for improvement. The key is how you prompt it.
Pre-Box 1: Diagnostic Prompt for Working Capital Health
First, read these files completely before responding:
[AR_aging_report_Q2_2026.csv] — Contains customer names, invoice dates, amounts due, and aging buckets (0-30, 31-60, 61-90, 90+ days)
[inventory_turnover_report.xlsx] — Contains SKU codes, category, quantity on hand, average days in inventory, and slow-moving flag (items with >90 days turnover)
[AP_payment_terms_summary.pdf] — Contains vendor names, standard payment terms, early payment discounts, and current days payable outstanding
Here is a reference for what I want to achieve:
[Upload a sample McKinsey-style working capital diagnostic report as markdown, or describe reference: “A one-page executive summary that ranks issues by cash impact, shows trend over last 3 months, and highlights quick wins under 30 days”]
Here’s what makes this reference work:
– Prioritizes issues by dollar impact (largest cash leakage first)
– Shows month-over-month trend for DSO, DIO, and DPO
– Identifies quick wins that require no system changes
– Uses clear, non-technical language for C-suite review
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Executive summary with 3 bottlenecks, each with root cause, cash impact estimate, and suggested action — 300 words max
Recipient’s reaction: Should immediately know which lever to pull first and see a clear link between data and recommendation
Does NOT sound like: Generic advice like “improve collections” or “reduce inventory” without specific dollar amounts and customer/SKU names
Success means: I can present this summary to the CFO in a 5-minute standup and they approve the action plan
My context file contains my standards, constraints, audience. Read it fully before starting.
DO NOT start executing yet. Ask clarifying questions first.
Give me your execution plan (5 steps max) before you begin.
What This Prompt Does
The diagnostic prompt forces ChatGPT to treat your data as a structured problem. By specifying the exact files, the reference format, and the success criteria, you eliminate vague outputs. The AI will ask clarifying questions—such as “What currency are the amounts in?” or “Do you want me to exclude intercompany transactions?”—before it starts. This step alone saves hours of back-and-forth. Once you answer, it produces a ranked list of bottlenecks. For example, it might tell you that “Customer X, with $450k in 90+ day AR, represents 22% of total overdue. Offering a 2% discount for immediate payment would recover $441k net of discount.” That’s actionable intelligence, not a generic suggestion.
Pre-Box 2: Tactical Improvement Plan Prompt
First, read these files completely before responding:
[working_capital_diagnostic_output.txt] — Contains the top 3 bottlenecks identified in the previous diagnostic session (customer AR aging, slow-moving inventory categories, and AP term renegotiation opportunities)
[company_payment_policy.docx] — Contains current payment terms for top 20 customers and top 10 vendors, including discount thresholds and late payment penalties
[cash_flow_forecast_next_quarter.xlsx] — Contains projected daily cash balances, debt covenant requirements, and minimum liquidity threshold of $2M
Here is a reference for what I want to achieve:
[Upload a sample turnaround playbook from a PE-backed company as markdown, or describe reference: “A week-by-week action plan with owner assigned, measurable KPI, and risk mitigation for each step”]
Here’s what makes this reference work:
– Every action has a single owner (e.g., “AR Clerk” not “Finance Team”)
– Each week has a measurable target (e.g., “Reduce DSO by 2 days” or “Collect $150k from top 5 overdue accounts”)
– Includes a contingency plan if the action fails (e.g., “If customer disputes, escalate to VP Sales within 48 hours”)
– Prioritizes actions that do not require IT or legal approval
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: 90-day plan with 12 weekly milestones, each with 2-3 concrete actions, owner, and success metric — 500 words max
Recipient’s reaction: The controller can assign tasks directly from the plan without additional interpretation
Does NOT sound like: A generic to-do list like “improve collections process” — every action must reference a specific customer, vendor, or SKU from the diagnostic
Success means: After 90 days, DSO drops from 52 to 45, DIO drops from 68 to 60, and DPO increases from 38 to 42, generating at least $1.2M in incremental cash
My context file contains my standards, constraints, audience. Read it fully before starting.
DO NOT start executing yet. Ask clarifying questions first.
Give me your execution plan (5 steps max) before you begin.
How to Use These Prompts Together
Run the diagnostic prompt first. It will give you the top three bottlenecks. Paste that output back into ChatGPT as the context for the tactical plan prompt. This creates a seamless workflow: diagnosis first, then execution. The tactical plan prompt also asks clarifying questions—for example, “Do you want to include weekends in the weekly milestones?” or “Should I assume the AR clerk works 20 hours per week on collections?” Answering these upfront ensures the plan is realistic for your team’s capacity. The result is a ready-to-use playbook that your controller can distribute on Monday morning.
Outro: Practical Next Steps
Start with the diagnostic prompt today. Pull your most recent AR aging report, inventory turnover data, and AP terms summary. Don’t worry if the data is messy—ChatGPT can handle CSV, Excel, and PDF formats. Upload them directly into the chat. The key is to be specific in your success criteria. If your CFO cares most about DSO reduction, say that. If inventory obsolescence is the bigger problem, adjust the prompt accordingly. After you get the diagnostic, run the tactical plan prompt. Within an hour, you’ll have a 90-day roadmap that is custom-fit to your company’s numbers. The only cost is your time and a ChatGPT subscription. The payoff? Hundreds of thousands—or millions—in freed-up cash.
Try it this week. Pick one bottleneck from the diagnostic and execute the first three weeks of the tactical plan. Measure the results. You’ll be surprised how fast working capital responds when you stop guessing and start prompting with structure.
Published on 9 July 2026 on growwithgpt.com