AI Series – Part 1: Identify a Business Problem
I decided to write this series on implementing AI as there are some great articles from many authors but I wanted to create a more hands on ‘actionable’ process with steps that anyone can start today. After years of working on data models and systems here is my take on the AI build. As a note I would suggest finding a local consultant that can support you in your journey. These professionals will provide you with direction and save you hours of pain to get you quick ROI and team user value. I would also suggest a team that has the full lifecycle available at their disposal. Someone who can create the design, implementation and do the development.
Before you begin take a few minutes to review the following articles:
This reviews the tech perspective of areas to focus in on projects and initiatives.
2) ‘Are your Designs Antifragile?‘
Reviews the mindset of building with Data as an Asset.
1) Identify a Business Problem or…
Depending on your business be it Small (3-50) or Medium (100-500) can determine the direction you will take for your AI journey. If you are a small business you will more than likely look to execute valued tactical needs that will involve plug and play, or ready-made AI solutions. These are AI tools that can be implemented with little risk with an immediate ROI focus. You might just implement a small AI app that saves an enormous amount of time for your employees to automating a process streamlining output. From the success of these wins you can move to a Medium model of AI execution.
The Medium (100-500) will most likely be focused on both the small wins of the ready made solutions but more strategic needs that will involve development of specific products designed for their specific business needs. There will need to be dedicated workshops, detailed roadmaps and it must be lead by leadership - all aboard the AI Train for a full transformation.
There will be a time when confidence is built to identify a full business lifecycle that drives value or revenue. An example would be a particular Sales, Marketing or Service lifecycle that has defined metrics of measure from start to finish and milestones of time to define where the actions and metrics are in the business process. From here you can identify one of these challenged processes (or even go another level down if needed) and begin.
Now some of these challenged processes will not work for AI so you may need a few to identify one that works. Once you do identify an opportunity, this process will allow you to measure the current state, future state and better define the Gap (gap analysis exercise) to find if you are moving toward success. It will be easy to see the impact to that process and surprisingly the impact to other dependent processes upstream and downstream. So, once you have defined a process, “Let the wild rumpus start!”
Where do we go next? Requirements and Use Cases!!! Read the following article on how to write clear and concise Requirements.
2) Data Analysis: