Working capital is where financial strategy meets operational reality. Every CFO and controller knows the uncomfortable truth: cash gets trapped in places nobody owns. DSO creeps up quarter after quarter. Inventory accumulates “just in case.” Payment terms get extended to appease sales, while suppliers quietly tighten their own. The result is a balance sheet that looks healthy on paper but feels tight in practice, with liquidity locked up in receivables, stock, and payables timing mismatches that no single team is accountable for.
The friction is rarely a lack of data. Most finance teams have ERP exports, aging reports, and cash conversion cycle metrics coming out of their ears. The problem is synthesis. Turning raw working capital data into a prioritized, defensible action plan requires cross-referencing receivables aging against customer concentration, mapping inventory turns against demand volatility, and negotiating payables terms without damaging supplier relationships. That analysis is slow, manual, and often gets deferred until a liquidity crunch forces the issue.
ChatGPT changes the economics of that analysis. Used correctly, it becomes a working capital analyst that ingests your context, applies structured reasoning to your actual numbers, and produces the kind of prioritized, board-ready recommendations that normally take a week of spreadsheet work. It does not replace judgment. It removes the blank-page problem and compresses the time between “we should look at this” and “here is the plan.”
Why prompting discipline matters here
Generic prompts produce generic advice. Ask ChatGPT “how do I improve working capital” and you get a textbook list: collect receivables faster, reduce inventory, extend payables. Useless. What a CFO needs is a prompt that forces the model to read your constraints, respect your industry dynamics, and output something a treasury team can actually execute on Monday morning.
The two prompts below follow a structured anatomy: task and success criteria, files to read, a reference pattern, a success brief, and an explicit instruction to ask clarifying questions before executing. That last step is what separates a usable output from a plausible-sounding hallucination.
First, read these files completely before responding:
working_capital_data.md — DSO, DPO, DIO, cash conversion cycle trends for the last 8 quarters, broken out by business unit
customer_aging.md — AR aging buckets, top 20 customers by outstanding balance, payment history notes
inventory_summary.md — inventory by category, turns, slow-moving and obsolete flags, safety stock assumptions
supplier_terms.md — current payment terms by supplier tier, early payment discount opportunities, critical supplier list
Here is a reference for what I want to achieve:
[Upload a prior working capital initiative memo or treasury action plan as markdown]
Here’s what makes this reference work:
It leads with the cash impact number, not the methodology. Each initiative is tied to an owner, a timeline, and a quantified cash release estimate. It separates quick wins (under 30 days) from structural changes (one to two quarters). It flags risks to customer and supplier relationships explicitly rather than burying them.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Prioritized action plan, 800-1200 words, organized by initiative with cash impact, owner, timeline, and risk rating
Recipient’s reaction: The CFO should be able to forward this to the treasury and sales ops leads without editing, and they should know exactly what to do first
Does NOT sound like: A consulting deck full of frameworks, or a generic list of “best practices” that ignores our actual numbers
Success means: At least three initiatives identified that together release a defensible cash amount, each with a named owner and a start date within 30 days
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.
Notice what the prompt does. It forces the model to ground every recommendation in a specific file. It demands a cash impact number. It requires the output to survive being forwarded without editing, which is a much higher bar than “sounds reasonable.” And it explicitly blocks the consulting-deck failure mode that makes most AI-generated finance content worthless.
From analysis to negotiation
The second prompt handles the part most working capital plans skip: the actual conversation. You can have the perfect plan on paper and still fail because a key customer refuses to move from net-60, or a critical supplier pushes back on extended terms. Negotiation prep is where finance teams lose weeks, drafting emails and rehearsing positions without a structured approach.
This prompt turns ChatGPT into a negotiation prep partner. It takes your customer or supplier context, your leverage points, and your walk-away position, and produces talking points, objection handling, and a recommended opening ask. It is the difference between walking into a call with a number and walking in with a strategy.
First, read these files completely before responding:
supplier_profile.md — supplier name, annual spend, contract terms, renewal date, criticality rating, alternative suppliers
payment_history.md — our payment behavior with this supplier over the last 12 months, any late payments, current terms
leverage_analysis.md — what we bring to the table: volume, growth trajectory, payment reliability, multi-year commitment potential
internal_constraints.md — our target DPO extension, maximum acceptable price concession, walk-away terms
Here is a reference for what I want to achieve:
[Upload a prior negotiation brief or supplier term sheet summary as markdown]
Here’s what makes this reference work:
It opens with the single strongest leverage point, not a summary of the relationship. Objections are pre-empted with specific counter-responses. It includes a clear opening ask, a target outcome, and a walk-away position. Tone is collaborative, not adversarial.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Negotiation brief, 600-900 words, with opening ask, target, walk-away, three likely objections with responses, and a one-paragraph email draft to initiate the conversation
Recipient’s reaction: The procurement lead should feel fully prepared to run the call without additional prep, and confident about where to hold firm
Does NOT sound like: A generic negotiation textbook, or a confrontational demand letter that risks the relationship
Success means: A brief that results in a term extension of at least [X] days at no more than [Y]% price impact, with the supplier relationship intact
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.
The pattern across both prompts is the same: constrain the model with your reality, define success in measurable terms, and force it to plan before it produces. Working capital optimization is not a knowledge problem. It is a prioritization and execution problem. ChatGPT will not tell you something about DSO that you do not already know. What it will do is take what you know, apply it to your specific numbers and relationships, and hand you a plan you can act on today.
Start with the first prompt. Feed it your actual last eight quarters of data, not estimates. Read the clarifying questions it asks before executing, because those questions will surface gaps in your own thinking. Then run the second prompt on your single largest supplier or customer exposure and see how much faster your prep cycle becomes. The compounding effect shows up in the second and third quarters, when the initiatives from the first plan are already releasing cash and funding the next round.
Published on 15 October 2026 on growwithgpt.com