Share this article

LinkedInTwitter
When it comes to artificial intelligence and cognitive computing, we couldn’t be living in more exciting times – new methods and technologies are invented every day, accelerating capabilities at a breathtaking pace. For a marketer, that means: what is completely impossible today, might be totally possible tomorrow, leading to higher uncertainty about business models and an even higher need to constantly innovate, especially when it comes to consumer products and services. This also means that the capability to innovate technologically is a constant necessity, with serious implications for investment cycles. For most mid-to-large size organizations an investment in a bigger building block of the corporate IT infrastructure would involve market screening for the technologies that best fit the current requirements, and a test and selection phase; before roll-out over a period of at least 6-12 months. However, as the AI technological innovation rate shortens, with significant advances every couple of months, these slower infrastructure investment processes cause business disadvantages. Not surprisingly, the solution is a componentized architecture, which allows IT services to quickly exchange technological capabilities or add them as completely new elements. For instance, a componentized architecture ensures that even if you couldn’t reliably discover logos and text in the video stream of your focus groups video interviews last month, you can easily add that capability tomorrow. Or if you couldn’t get answers to questions about a consumer segment across all your categories and structured and unstructured datasets today, you can probably do it tomorrow – just add a meta-domain model as a module, without having to exchange the whole platform. For such a componentized architecture to function, it’s essential to clearly separate functions and data. From an architectural standpoint, that means separating the functional layer from data ingestion over transformation and normalization, persistence, analytics, and aggregate functions to applications (phew). It’s also crucial for each layer to be built with modularization in mind, so technical processes and frameworks allow for easy exchange and addition of components. At Market Logic, our insights platform has been consistently built with a componentized architecture in mind because we want to deliver the best possible, and always up-to-date solutions for our customers’ marketing challenges and tasks. And that means we won’t always be the inventor nor the creator of all required technologies to fulfil that goal. So when LivingLens develops a great image analyzer, we plug it. When Google introduces a great speech to text engine, we’ve got it. When RaRe Technologies release a document similarity algorithm, we love it. Using an open source UIMA componentized architecture for unstructured information management gives us the flexibility to quickly adopt technical and functional capabilities developed by the best creators in the business, where a proprietary platform would leave your marketers waiting months or years to capitalize on the latest breakthrough.