Artificial intelligence (AI) is not just a buzzword, it’s a megatrend. Half of the technologies discussed in Gartner’s "Hype Cycle for Emerging Technologies 2019" relate to AI. Practically all companies, whether large or small, stand to benefit from the countless opportunities presented by AI to improve, or even transform, their businesses. However, many people are unaware that large swathes of AI development are available to everyone. Through our “data science in the cloud” offering, our experienced team of business consultants, data/software engineers and data scientists support companies throughout their agile and scalable AI journey.
Standard or customised
Turnkey AI products such as cloud-based image and speech recognition (applied AI) packages are ideal for automating specific processes. But such products alone are not sufficient to deliver fundamental competitive advantage and additional business benefits. The first step is to pinpoint where in a company’s business processes data is already available that could be utilised for custom AI solutions, and where it would be possible, or sensible, to collect new data. Use cases for the deployment of AI (custom AI) can be derived directly from a company’s strategy; What problems can we solve with AI-based data? How can we monetise existing data? What kind of data-based products or business models could be devised?
At Ergon, we take the following iterative approach in order to move from a given business idea to a workable AI solution:
- Potential analysis
First, we check ideas for their technical and data-specific potential. Simultaneously, initial rough data analyses are performed – ideally in the correct AI pipeline, which can later be scaled.
- Data engineering
Data forms the basis for training AI algorithms. Common stumbling blocks include varying data quality or non-combinable data obtained from different silos. Data engineering seeks to address these challenges from both a technical/data-specific and a specialist/organisational perspective.
- Data science
A core part of the process, data science applies appropriate techniques and approaches to extract the relevant technical information for the AI models from existing data.
- Software engineering
Here, the successful AI algorithms built during the data science phase are translated into a robust, scalable solution.
Applications of custom AI
Artificial intelligence has become indispensable in some areas but there is still plenty of untapped potential in others:
- Unthinkable without AI: automated fraud detection solutions
- IoT + ML: Optimisation of networked production plants – e.g. for predictive maintenance
- AI-supported prediction models for planning orders and production processes – e.g. replenishment in retail settings or planning raw material requirements in a manufacturing context
So how should companies proceed? They should choose an approach that enables both fast, low-cost potential analysis and direct, clean-data engineering/data science. This should then be followed by professional and secure implementation of a productive AI solution. Thanks to data science in the cloud, barriers to entry have fallen and become easy-to-deploy. Sustainable options are now widely available for organisations seeking a successful start to their custom AI journey.