Planning and implementing an AI solution can be very challenging. If it’s the first time you’ve ever dealt with this kind of project, you might even find it intimidating. AI is a relatively new domain, and AI systems are very different from your average traditional system. While machine learning algorithms are at the heart of the solution and require special expertise, implementing them properly in an organizational production environment entails close consideration of both the technology and software environments.
To better understand and prepare the design of your AI solutions, we highly recommend that you take the following preliminary steps:
Mapping and examining relevant data sources The main data sources for an AI solution are usually fairly straightforward, but we still need to think about mapping the relevant pieces of data and be aware of the challenges of retrieving and connecting them. Also, some data sources are not always obvious, even though the data might be relevant, so we should recognize them and keep in mind their potential significance. We should also think about external data from vendors or even from public data sources that might be relevant to our project. Using external data can be challenging, so it might be a good idea to deal with it in the early stages.
Understanding architectural guidelines and constraints Although this is not the right time for detailed technical design, we will eventually need to decide and know exactly who is running what, and where they’re running it. Even now, before diving into the details, we can identify infrastructure and systems that we know will be involved and others that we should keep in mind because they might be relevant. We should also think about new components (software, hardware, cloud services...) that we might need to purchase. It will also be helpful to map components we know that we ourselves will need to develop from scratch. Mapping these insights to the general architecture can be extremely helpful for the next steps. We can then start to update the diagram with actual components and attributes.
Identifying and engaging project members and Stakeholders Knowing the general process and components of the system, we should be able to identify who would be relevant members of the project team and the roles they would perform. Contacting them and getting them interested can be very helpful. We might also discover a knowledge gap in human resources. This information could easily be critical for project planning and might affect future project solutions and decisions.
Reduce knowledge gaps AI is a complex and challenging area, and planning the right working solution is even more challenging. Knowing the components and process and the overall picture should highlight any knowledge gaps we might have. We should “dive in” and get a better understanding of the technologies involved, and then we can examine and test different products and solutions.