Large language models have made some work cheap that used to be expensive: drafting text, summarising documents and answering questions from a body of knowledge. For a business owner with a method, that opens up new things a product can do. People still pay for the result the product gets them.
Where AI helps
AI is most useful where your method already involves reading, writing or answering questions. In products we have built, it drafts operating procedures from a franchisor’s own documents in Manual Magic, answers questions from the finished manual, and helps authors move from an outline to chapters in Book Magic. In Wowment it flags conversations that may need a safeguarding review.
In each case the AI works inside a method someone already trusted. The method decides what good looks like, and the AI does the drafting.
Where it needs care
AI can be confident and wrong. Anything a customer relies on, such as advice, figures or safeguarding, needs a person or a clear check before it reaches them.
It also costs money each time it runs. Price the product with that in mind, and use AI where it saves the customer real time.
Then there is data. Customers will ask where their information goes. Decide before launch which model providers you use, what they can keep and whether anything is used for training. In the products we build, nothing a client gives us is used to train anyone’s model.
How to add it
Start with one task your customers find slow and your method already describes well. Build it, measure whether people use it and whether it saves them time, then decide on the next one. One AI feature that customers can rely on is worth more to them than five they cannot.
Next steps
If you are deciding where AI belongs in your product, request an invitation and tell us about your method.