CONSIDERATIONS TO A SUCCESSFUL AI STRATEGY
As a first step to defining your AI strategy, we need to create our business strategy. This entails understanding our stakeholders, our mission and vision, our priorities, key pain points, etc. Based on this we need to assess our current state and move to the next stage.
Most AI/ML solutions need data to provide actionable insights. At this stage we would like to understand the data we need, the current data maturity, volume of data, data storage, proprietary/ open-source or third party data. The ideal scenario would be able to set up a automated data pipeline which captured all needed data into secure, accessible and scalable suitable data stores.
As a next stage we need to identify the technological infrastructure we would need. The consideration would involve the data source, the current state of technology, the ML Procedure, the implementation, short term and long term goals. The outcome at this stage will be the choice of cloud infrastructure, security of the system, scalability and the most important open source vs. proprietary.
In the implementation phase we need to understand who are the consumers of our data and AI deployment. We need to decide if the results need to be streaming for example in trading or it needs to be in request-reply mode say a churn model for a or can be delivered in batch mode for example a financial recommendation system for UHNI. Another consideration is whether this will be consumed in a web based app or mobile app. The concurrent usage of the results is yet another driving factor behind implementation considerations