The ML Journey doesn’t have to be Daunting!
by; Geoff Craig, Director, Solution Services
Beginning the journey to Machine Learning (ML) assisting and executing business process can seem like a daunting task. A recent article by Gartner shows that while the adoption of artificial intelligence (AI) grew by 270 percent in the past four years, the total percentage of companies using AI is still only 37 percent. Given the amount of changes in AI technologies and the amount of media hype around AI most companies are trying to work out how AI can help transform their business. Once you start digging into what implementing AI into a business process really means, you start finding out that implementing AI into your company’s IT stack is just like implementing any new technology. There are the core questions; what new hardware and software will I need, who will manage the system, who will use the system, and is a hosted solution the way to go? Along with these standard questions there are a new set of questions like which type of model should I use, what data do I need to build a model, where will that data need to live to train a model, how will the data be secured? Then the final (and most interesting for myself) is how do I serve the model, at what step in the business process do I insert the model and how does the model change the business process?
A model that has been tweaked and tuned to provide the best predictions is useless if it is never used to serve predictions, so a CS Team works with a customer to ensure that a model can be served and used within a business process. Introducing models into a business process can involve multiple teams from multiple areas of the business. This is where guidance and leadership is sometimes needed to introduce and explain Machine Learning technologies to both technical and non-technical leadership.
As you can see starting the journey to using ML in your business can seem daunting and while there are numerous resources available, from blog posts, public repositories of ML models to tons of documentation, it is still hard to put these disjointed pieces together to create a complete ML solution. This was a problem that we wanted to tackle early on, so as we develop and continuously improve our AI platform we decided to ensure that our customers and partners had the resources available to tackle this problem head on by creating our Jumpstart and Solution Accelerators Programs. The goals of both programs are the same, to help customers and partners create quality AI solutions in the fastest manner possible. The difference is Jumpstarts are an engagement between customers and Lucd and our Solution Accelerators are more of a step by step guide showing common ML solutions that are applicable to numerous verticals.
To enable your company to become an AI company we suggest you start small, start quick, get results and then scale. This thinking closely aligns with Andrew Ng’s AI transformational Playbook, both allowing for faster adoption of AI via repeatable models. Your journey to ML will help you organize your company for better decisions and can lead to new opportunities as well as potential new business innovation and finally the ability to monetize your stored data.