How to start with AI?
We explain some clear steps to undertake before starting an AI project
In the last couple of years Artificial Intelligence has drastically moved its applications to more and more industries. Regardless of a company's size or type, its executives typically look for ways to help its organization in operating as efficiently as possible.
Many executives hear about Artificial Intelligence and the massive growth it can deliver within an organization. When considering an AI-based project, the questions which executives often ask are: 1. Where to start, 2. Do we have enough data, and 3. Who should we bring around the table to decide on possible AI use cases.
Knowing the Areas of the Business that you Want to Improve
If you are a business that has a clear plan of the challenges that it is facing and the objectives that you want to achieve, your organization may be ready to implement AI. There is almost an unlimited number of business challenges that machine learning and predictive analytics can solve for your business. Harvard Business Review estimated that AI will add $13 trillion to the global economy over the next decade. Hence why you should have a clear idea of what are the most painful challenges to you, and if AI can help you solve them. Even if you are a small business, operating B2B or B2C, there are many areas where AI can help you gain competitive advantage and increase market share. AI can help and reshape businesses in areas like organization and management to process automation.
Have an Overall View of The Data
Data is an important aspect in most AI projects. Many AI techniques are based on having a lot of data that is used to train the algorithm, allowing the model to operate over new data.
In order to understand if and how AI can help you, it is useful to have an overview of the data sources that are held by your company. Having a list of the data sources, what they contain, regarding which time frame, is important to understand what you can do with it. The more detailed the view, the more you will be able to have a qualitative feedback from an AI expert such as the Kantify team.
Having sufficient data
The data needed to create a custom-built AI tool may differ depending on the nature and complexity of the model. We have divided different AI use-cases into three categories, depending on the amount of data needed for the model to perform well:
- more than 1.000 data examples for: Customer clustering, Sentiment analysis, and Image classification;
- around 10.000 data examples for: Hyper-personalization, Anomaly Detection, Churn Detection, and Lifetime value prediction;
- more than 100.000 data examples for: Predicting Complex Customer Behaviour, Predicting market trends, and Extracting data from unstructured sources.
This is just a high-level view, that differs depending on the use case. So contact us if you have a specific use case in mind so we can provide you a detailed assessment. For some companies, a lack of data can lead to difficulty in implementing an AI-based solution. For that reason, we help companies organize, process, and examine their data, also point to external datasets that can obtain insights.
Know Which Stakeholders Can Support You
Like any new project, AI has to be considered in terms of change management. You are going to introduce a new technology and need the right stakeholders to support you and commit the necessary support, or budget, or IT resources, or time. This is why you should know what are your key stakeholders so they can be onboarded when time comes. Again, this is something in which we also have experience, so ask us about it!
It is essential that businesses invest in implementing AI-based solution in the area that is most needed. Identifying the overall business goals and objectives in the short, medium and long run is a good point to begin with, and the next step is assessing technologies that can help achieve those goals. Depending on the company's challenges, different use cases can be applied. You can read more about different use cases for AI within a company here, and/or book a 15min slot with one of our team members to discuss your questions (on English, French, or Dutch).