Machine learning and artificial intelligence… It sounds cool, but how can you do that without having a datacenter full of data and a team of data scientists? At Kunlabora we learned that with the 2020’s in sight the tools are there to start incorporating artificial intelligence and machine learning in any business. And all this without requiring huge investments and lots of time, but gradually and step-by-step.
“No step forward is too small. Just be sure it’s taking you to the right dream, then take more of those tiny steps.”
By Israelmore Ayivor
Over the past decades companies like Amazon and Google gathered a tremendous amount of data, which enabled them to develop a new kind of services that use machine learning technologies to do more with all this data. Netflix, Youtube, Siri, Google Assistent, webshops, … We all get in touch with machine learning algorithms and artifical intelligence through their services on a daily basis. And this new type of services helped them create extra business value by differentiating their offering. Now the cloud companies are offering these technologies as part of their cloud services. Things like automatic translation, speech recognition, computer vision, chatbots are not just for the big companies anymore, but their research and experience resulted in the currently offered services. This makes them available and affordable for every company.
But how does this help you in your business? Well, what you certainly need to know is that whether your company is big or small, profit-driven or impact-driven, it doesn’t matter. If you have a need where this modern technology seems to offer business value the tools are there to integrate in your solution and to start learning.
In our Mezuri project we use a machine learning algorithm to do sentiment analysis on short pieces of text. The sentiment feedback is one of the elements that help reporting on how impactful the work is that is done by family coaches. By using an existing machine learning algorithm we didn’t need to do a lot of research upfront that would need a lot of data and a team of data specialists. On the contrary our users immediately get extra feedback on their recordings and learn about possible improvements.
In our Repair App project we integrated machine learning algorithms for computer vision to identify electronic devices. In the first version of the app we started with using existing machine learning algorithms for OCR (optical character recognition) and barcode reading, knowing these algorithms are not optimized yet for this specific use case. The challenge in this project was to build an application that uses computer vision technologies that can be replaced by newer and more refined algorithms later. Having an application ready to illustrate the possibilities of these technologies adds early value since it facilitates feedback from real users and helps to get a better idea of the extra data and business logic that is needed in a real business context. By the way, did you know that we even are running this application on pay-per-use infrastructure, meaning we can keep the running costs extremely low?
We do not have huge amounts of data. We are not a team of researchers that is developing machine learning algorithms. We are not a team of pure data scientists that is using data to train models and implement this artificial intelligence for specific use cases. We are a team of software developers with a genuine interest in machine learning, artifical intelligence and data engineering. We do have knowledge of machine learning algorithms and know when to use them and how they can bring you extra business value. We help you to find and build a solution for your needs, and if these modern technologies seem to add real value for you, we will integrate them in that solution.
“What’s dangerous is not to evolve”
By Jeff Bezos - CEO Amazon.com