Tax Tip of the Week: We Asked AI to Draft a Business Plan. Here's What We Got.

"The best way to cheer yourself up is to try to cheer somebody else up" 
     - Mark Twain

Here is an article from the Wall Street Journal that discusses AI's strengths and weaknesses when writing a business plan.

You have a great idea for a new business. Is it time to turn to AI?

We decided to find out by asking two different generative-AI tools to write a business plan for a fictional startup—a travel app focused on the U.K.’s Yorkshire region. What did these tools do well, we wanted to know, and what are their limitations?

To get started, we provided the tools with a detailed prompt that specified what we wanted in the business plan. We also gave the tools permission to ask follow-up questions if they needed more information and to be creative if there were gaps in their knowledge.

What did we find? In short: The business plans produced by AI were useful, but not quite ready for prime time. That is, they could serve as a good starting point for a company founder, but are in need of further editing and additions to pass scrutiny. Not bad considering where we are in AI’s development.

Here’s a closer look at four things the AI tools did well—and four shortcomings:

AI’s strengths

1. Structure and logic: Perhaps not surprisingly, generative AI excels at following a logical structure. The business plans we received adhered to the typical format and contained all of the sections we expected to see, including an executive summary, followed by a company description and then a detailed value proposition. This structure provided a solid foundation for a business plan, making it easier to expand and further customize the document.

2. Filling in some details: AI was able to fill out each section of the business plan with at least some relevant information that wasn’t provided in the original prompt. That included information about the company’s mission, market strategy, product offerings and target market.

“Yorkshire Tales is a great way for history enthusiasts to learn more about local history,” AI wrote. AI even suggested a plausible amount required from investors—“Yorkshire Tales is seeking $1 million in seed funding to launch its business”—though it mistakenly used the U.S. dollar symbol instead of the pound sign. Still, much of the information provided by AI would have significantly reduced the time and effort needed to produce a first draft of a business plan from scratch.

3. Creativity (up to a point) : AI isn’t going to win any awards for creativity. But it was able to produce catchy names and concepts that made the business plan more engaging. For example, AI named our company “Yorkshire Tales,” a simple play on words that tied together the proposed focus of the business with its geographical origin in the Yorkshire Dales, a part of England known for its landscapes, history and rich storytelling traditions.

4. Highlighting key elements: AI also was good at identifying and highlighting the critical elements of a business, such as its unique selling points, target market and competitive advantages. For example, AI pointed to the unique approach offered by Yorkshire Tales in combining local history with modern technology.

AI’s weaknesses

1. Lack of detail and analysis : While the AI tools did offer some details, they still fell short in that department, especially regarding market data, financial projections and unique business challenges. Perhaps an even more detailed prompt would help address this, but at some point, it might be just as easy to do it yourself.

Similarly, we found generative AI’s ability to conduct a thorough market and competitive analysis to be limited. AI currently lacks the ability to learn directly from real-time data and isn’t able to provide in-depth, nuanced analysis of market trends, competitor strategies or customer behavior as would be expected in a business plan. A human would have to get involved for that, at least for the time being.

2. Repetition: The text in our AI-crafted business plans tended to repeat certain phrases and concepts—to the point of being boring and distracting for potential investors. For example, phrases such as “Yorkshire’s rich heritage” and “the business’s deep understanding of Yorkshire’s history and culture” were used repeatedly. A prompt that discouraged repetition could help, as could requests for AI to write in a compelling style.

3. Occasional inaccuracies: Generative AI is famous for its “hallucinations”—facts that are made up and, well, not always factual. Such inaccuracies, if not spotted and corrected, would undermine the credibility of a business plan. One glaring example in our case was AI incorrectly using the U.S. dollar symbol, rather than the pound symbol, when describing our fictional company’s seed funding, a detail that could confuse or mislead readers and deter potential investors.

4. Lack of personalization: Generative AI was able to produce a basic business plan, but without further prompting it struggled to personalize or adapt the structure to the business being described. That resulted in statements such as, “Initial funding will be sought from local investors and grants focused on promoting local culture and history” and, “Revenue from app sales, in-app purchases, and partnerships will be reinvested into content creation, marketing, and app development” for the financial section of the plan. As with so much of the AI-generated business plan, it wasn’t a bad start—but the technology wasn’t yet ready to produce something that would win over investors.

Credit goes to Gordon Fletcher and Maria Kutar, published in Wall Street Journal January 19, 2024
 
Thank you for all of your questions, comments and suggestions for future topics. As always, they are much appreciated. We also welcome and appreciate anyone who wishes to write a Tax Tip of the Week for our consideration. We may be reached in our Dayton office at 937-436-3133 or in our Xenia office at 937-372-3504. Or, visit our website.
 
This Week’s Author, Mark Bradstreet

-until next week