AI for Small Businesses: Ideation
In the third post of our blog series "AI for Small Businesses," we address a use case not immediately associated with GenAI: generating and developing ideas in dialogue with a chatbot.
As in the previous posts (Part 1 and Part 2), we specifically address small businesses looking to test GenAI's capabilities in daily operations without investment. For this reason, we limit ourselves to generalist GenAI chatbots offered as a service.
Caution! When using chatbot services, adhere to two principles:
- Never disclose confidential or sensitive information to chatbots. Your inputs could end up in training data or evaluations and be accessed by other users.
- Check GenAI texts for their plausibility and appropriateness. GenAI chatbots are not subject matter experts. They make up things that neither match reality nor your requirements.
For the examples in this post, we will use Claude. Claude offers a free version that you can test after registration. All the examples we describe were tested with the free version. Recreate our use cases with the shown prompts or experiment with this tool yourself to see if it can help you in your daily work!
Critically Examining an Idea
As described in Part 2 of our series, a GenAI chatbot can take on various roles with appropriate prompts by defining the context and tone. In this post, we will use this to create feedback from suitable perspectives.
We will use Claude to get feedback on a new business idea on behalf of our fictional company "GreenGrow." We use the following prompt:
Here is the description of a fictional company: <description>Imagine a small company called "GreenGrow" specializing in sustainable horticultural products. The five-member team works in a cozy downtown office surrounded by plants and eco-friendly gadgets. Their days are filled with customer consultations, maintaining online presences, and collaborating with local suppliers to offer a wide range of organic fertilizers and recyclable gardening tools. The open office atmosphere fosters creativity and direct communication, with team members often flexibly acting in various roles.</description>
Take the role of a sales expert in horticultural products acting as a mentor for the young company GreenGrow. I am the CEO of GreenGrow and have an idea for a new product. Give me critical feedback on my idea. Formulate your feedback as a list. Limit each criticism to 50 words. Here is my idea: <idea>"Seed & Save" is a subscription service that combines seasonal seed packages with detailed instructions and eco-friendly gardening accessories. The goal is to promote urban gardening and bring sustainability directly to urban households.</idea>
GenAI chatbots can interpret text markers or delimiters in prompts. In the case of Claude, this marking happens as demonstrated above using "tags"
Claude respondes to our query with a list of eight critical factors that we should consider further; such as: defining the target audience, thinking about logistics, setting us apart from competition, scalability, etc.
We could then choose to delve deeper into individual points on this list in a conversation with Claude. We might, for example, ask the chatbot: "Expand on point 5. What strategies can we use to achieve customer retention? Give me five examples."
It is important to note that GenAI chatbots are good at giving the impression of being experts in a field. This is sufficient to ask critical questions that allow an idea to be developed further. However, this feedback is not equivalent to genuine subject matter expertise. In particular, we advise against using GenAI chatbots to check legal documents or other critical texts.
Applying Specific Conversation Methods to Develop an Idea
Our next use case is closely related to the critical feedback described above; but makes even more use of the conversational nature of chatbots.
Just as modern GenAI chatbots can mimic various roles, they can also apply any discourse patterns to critically examine ideas. For common practices, you can assume that the chatbot can act without further instructions (just ask if a method is known); in other cases, you may need to explain the principle beforehand.
In the following example, we use the method of the "Socratic dialogue"
Apply the Socratic method to help me define the product in more detail and further develop the product idea.
Even without further specification, Claude formulated targeted questions that aimed at critically examing our new product. He asked how we define "urban gardening", digging further into our definition and making us re-think our target audience and how the product would be applied; before doing the same for other aspects, such as how we would provide "eco-friendly" products.
In this way, we gradually worked through some critical aspects of product specification. We concluded the process by instructing Claude to provide us with a summary of the information gathered in the form of a product description.
Re-framing an Idea for Presentation
Finally, we want to write a news post about the new product on behalf of GreenGrow. However, the information is unstructured and mainly in keywords or bullet points. To now formulate an ordered, easily accessible text, we instruct the chatbot to take on the role of a journalist and ask questions about "Seed & Save" until he has enough information to write an article. We then use the result as a basis for our news post.
I want to write a news post about "Seed & Save" to announce and introduce the product to our customers. Take on the role of a journalist writing a post about "Seed & Save." Ask me questions about this product until you have enough information for your post. Once you have enough information, make a suggestion for an appropriate news post. The post should be positively worded and make the product attractive to excite customers about it. The post should be no longer than 200 words.
Claude responds with a single, structured list of questions. Depending on your preference, you can also instruct the chatbot in the prompt to ask the questions one by one instead of all at once. This results in a different dialogue with different outcomes. After we answered Claude's questions, he made the following suggestion for a news post:
Thus, Claude helped us structure the new idea and shape it into a form that we could use to communicate the concept to customers or investors.
When using this method, we recommend that you review the text before publishing it according to your discretion and needs, as it may contain factual inaccuracies as well as opinions you do not wish to publish in your name. The tone of the text usually do not match the company's identity, making it evident that you did not write the text.
Abschluss und Ausblick
In the third post of our series "AI for Small Businesses," we explored how generalist GenAI chatbots can support idea generation and structuring. We hope that these three use cases with Claude have given you some new input on how you can use GenAI to develop your ideas; without making any investment. Always keep our two behavioral principles when dealing with chatbots in mind.
In the next post, we will address the topic of AI-assisted search and research – see you soon!