Making AI work for you: Advanced prompting strategies for B2B credit leaders

Artificial intelligence (AI) has taken a life of its own, integrating itself into almost every facet of our daily lives. Over the years, its use has only increased in both professional and personal settings. However, AI is only as useful as the information you provide it, and without the right context, it can generate false or inaccurate responses. 

Why it matters: Credit professionals who learn how to best leverage AI are the ones who will stay ahead. By using effective prompting strategies, they can tailor it to deliver more accurate and actionable results. 

How Credit Professionals Are Using AI 
What started as a tool to research customers and draft letters has evolved into a means of streamlining various credit functions. “I use an enterprise version of Copilot for administrative tasks, like drafting agendas and summarizing meetings, and for credit-specific tasks, like reviewing and updating our credit policy,” said Haley Hagin Fimbres, CBA, CICP, credit manager at Superfeet Worldwide LLC (Ferndale, WA). “I also use it to revise collection and demand letters for tone, ensuring my message remains clear and professional.” 

Neural networks can instantly organize, review and generate large volumes of data, a feature that’s proved advantageous for credit professionals. “I present financial documents to AI—absent of customer-identifying information—and direct it to calculate specific ratios, summarize sections, pinpoint up or downtrends and consider the credit request (by amount, terms and industry),” said Krissy Aeschbacher, credit analyst at Maverik Country Stores, Inc. (Salt Lake City, UT).  

Some credit professionals use intelligence systems for credit approvals or establishing credit limits for customers. “I would input the appropriate credit application fields along with the corresponding credit report of the customer to evaluate risk level,” said Hagin Fimbres. “We’ve refined the process over time so it can discern accounts according to our company’s risk tolerance.” 

AI-driven contract analysis tools can automate the review process, swiftly identify key terms, potential risks and anomalies within contracts. “I ask Copilot to review contracts for unfavorable terms or missing information and to check if there is a mutual indemnification clause,” said Hagin Fimbres. 

However efficient generative AI can be, it’s not a perfect system. Its tendency to produce false or inaccurate information makes it a liability at times. Refining prompting strategies will help you receive the most reliable AI assistance possible. 

Optimizing Your Prompts: Specify and Over-inform 
Being specific and disclosing information in your prompts will help you get exactly what you want, when and how you want it. Providing details about your role, company, industry and customer can provide sufficient context for more intuitive answers. “I’ve told AI to write me a professional but semi-firm collection letter for a customer that is X number of days past due that I want to retain in the B2B world,” said Sheila Roames, CCE, western regional credit manager at Ergon Asphalt & Emulsions (Chandler, AZ). “I tell it I’m in the asphalt industry and I’m looking for A, B and C. In the letter, I want to reference prior reminders and request payment within X number of days. I’ve found that providing this type of context as well as my desired outcome helps generate relevant content.” 

Telling your AI platform what you don’t want can help enhance results and save time from making multiple revisions. For example, you can end your prompt by saying, “Avoid this” or “Do not include this.” You can even go a step further and preface your prompts with, “For future reference,” “From now on” or “Moving forward,” to guide future responses. 

Use Examples 
Using an example or a template can give the AI a reference point to help improve results. For example, if you’d like it to emulate a credit application, just say, “Follow the structure of this example,” “Draw from this,” or “Using the X attached.” However, it’s very important to not infringe on copyright. Don’t upload copyrighted works and ask the AI to rewrite/reproduce them or create a substantially similar work. 

Provide and Ask for Feedback 
Providing and asking for feedback will enhance the software’s findings and help you formulate better prompts. For example, you can ask for advice on how to get to a desired output, including what information is most useful and what prompts would be most effective. For instance, you can say, “Critique my prompt from the perspective of producing high-quality credit risk outcome,” followed by, “Be specific about what’s vague or missing.” 

Your guidance will enable AI to come up with better solutions that align with your needs. “It helps to treat it like a conversation,” said Aeschbacher. “I’ll write something like, ‘This is helpful, but it’s not quite right. Please remove X, adjust the tone to Y and expand on Z.’ That kind of iteration always gets the document closer to our company’s voice.” 

Treat it Like a Human 
One way to generate more compelling outputs is to prompt AI as though it were a person, which in this case, would be a credit professional. Instead of asking, “What are the risks of extending credit to a new customer?” try, “If you were a seasoned credit manager with 20 years of experience in the manufacturing industry, what red flags would you look for before extending credit to a new customer?” By assigning a role, you give the AI a framework for tone, perspective and expertise to produce responses that feel more grounded, specific and human. 

The bottom line: Prompting AI takes practice, but the investment is worth it. Credit professionals who treat AI as a collaborative tool, rather than a simple search engine, will find it capable of handling everything from drafting collection letters to analyzing financial statements.