Where to start in AI is confusing for everyone, ISV and client. There are too many marketing messages for non-AI practitioners to understand what is truly happening.
We recently started working with an ISV that provides its clients with a training and education platform. The ISV offers a specialized application to help small and medium-size clients (SME). Our client relies on a large amount of consulting revenue on top of their application license and maintenance fees.
Like most ISVs, they are looking for recurring revenue. They liked the consulting revenue for that but look in the future to move their software to a monthly fee without upsetting the existing client relationships. We couldn’t help them with their main software but we did design a recurring revenue product using AI and predictive analytics that fit into their offering portfolio.
We took our market discovery work of what their clients wanted to see for AI and predictive analytics to design a subscription service offering that blended with their go-to-market strategy. The service designed emphasized the outcome of the predictions delivered to the point of use of managers whose employees were looking at training information and the HR department responsible for the overall application. We focused on just a handful of high value, high impact prediction models leaving a long list of potential areas for AI in the future.
Working with their development team is going well as is the first set of functions we are testing. The total delivery of the minimally viable product won’t take long so we are beginning the work of creating marketing materials and sales processes they will use.
The AI and predictive analytics development process does not need to take a long time. The key needs to be working in parallel between departments. The worst situation is to wait for the development to be “perfect” before engaging marketing and sales. Think of this as agile development with the involvement of development, support, marketing, sales and a select customer or two.
This is true of almost any software product but is doubly true for AI. AI is probability-based – there is no certainty only an acceptable margin of error in making predictions. Trying to squeeze out an extra percent or two of accuracy can easily triple the time spent developing the AI models. There is time to do that after getting real-world experience with clients.
If ISVs wait too long they will find their customers are already engaging standalone services or small, specialist consultants. Start the conversation now – deliver something small but usable soon – and then enhance the offering over time.