What is Visual Intelligence?

Visual Intelligence is a category of enterprise software that harnesses advanced artificial intelligence techniques applied to visual data to drive digital transformation. Developing and deploying Visual Intelligence at scale requires a new technology stack.

The cornerstone for digital transformation

AI, machine learning and data science applied to an expertise domain are strategic skills for a n  enterprise in digital transformation. It is part of the vertical knowledge and expertise that industry leaders must master as part of their core business.

The complex underlying technology stack, necessary to enhance such digital skills, is complex to handle and too far from the core business. Furthermore, it can advantageously be mutualized between verticals to lower operating costs through economy of scale.

For industry leaders, the natural strategic approach consists of choosing the right Visual Intelligence technology stack, developed, maintained and operated as a service by a third-party specialist. As it will lay down the basis for their company-wide visual data management and Information System, they can focus on their domain of expertise and maximize their value creation through the quick and efficient development of vertical-specific AI applications.

Enterprise software innovation cycles follow a typical pattern. Early in the cycle, companies tend to adopt a “do it yourself approach” and develop technologies in-house. It is a virtuous phase, as it is part of the learning curve. It raises the consciousness of the problems and challenges to overcome. Then comes a rationalization phase, during which the work is divided between the “generic” that is productized by enterprise software companies, and the “specific” that is internalized by companies, sometimes helped by third party system integrators. For each particular end user, the optimal solution becomes the appropriate combination of both.

These cycles have reproduced over the years 1990 and 2000 with ERPs, then CRMs, with results we see today. The current cycle  is Visual Intelligence, and more generally enterprise AI. The complexity of a Visual Intelligence stack is 2 to 3 orders of magnitude  higher than the one of a CRM. The same principles apply, only multiplied.

The choice of the right Visual Intelligence stack for an organization is a challenging one. Many CIOs are currently in the process of doing it. One of the important cursors to position is simplicity vs versatility. Simple means easy and economic to implement, but with few customization options and a limited scope of work. Versatile means adaptable, customizable and future proof, but comes with a higher degree of complexity to handle.

A typical simple configuration can be the choice of one or several best of breed SaaS applications that can cover most or even all the needs of the organization. They will be very simple to deploy and to use, no maintenance or development is required. They should be economical. They will realize simple, laser-focused tasks, very efficiently. However, little configuration options are available, with little to no integration possibilities. The organization somehow has to adapt its processes to the tools.

A versatile configuration is typically the adoption of a Platform as a Service (PaaS) that will be the underlying technology stack for the creation of an integrated and optimized Visual Information System. While the platform is maintained and continuously improved by the provider, a workforce is necessary to perform the integration tasks and the development and maintenance of specific business applications. Depending on the strategic positioning and the size of the project, these tasks can be carried  out by an in-house team of IT developers and data scientists or a third party system integrator (or a combination of both). This solution will yield a fit-for-purpose and integrated information system. It is future-proof as it is designed to be evolutive and compatible with an important paradigm: start aggregating data now to capitalize on it later. Create the conditions to find value along the way.