Intelligence needs thinking about

Building an A.I strategy often starts with the basics. (Even Skynet wasn't built in a day)..

A.I Aspirations

Companies of all sizes are exploring options for incorporating artificial intelligence into their business processes. Potential use cases are vast, and include deployments that improve efficiency and productivity, facilitate better and faster decision-making, uncover relationships in large data sets, and provide more personal, contextualized customer interactions. In addition to horizontal applications, such as those that manage sentiment, monitor security, or perform sales related projections, AI is also making its way into industry-specific applications. It is being used by pharmaceutical companies to speed drug development, by the financial community to review loan documentation, and by retailers to suggest product purchases.  AI deployments range in complexity depending on the details of a particular project or use case. Organisations looking to get started with AI by incorporating natural language generation into an existing application, or by launching a pre-built chatbot, will have very different requirements than a more ambitious team that is interested in creating their own customized machine learning or deep learning model. Not only is building, training, tuning, and managing an AI model a daunting task, but so is identifying and cataloguing data inputs, particularly vast amounts of unstructured data. Enterprise AI platforms streamline the process by providing an integrated environment that facilitates the development for AI solutions, for a range of audiences. Pre-built solutions are more readily accessible to non-AI specialists, and integrated toolkits, libraries, and resources help streamline deployments for data scientists.

Market Definition

Cloud-based enterprise AI platforms consist of tools and resources that allow enterprises to incorporate cognitive functions into their applications and business processes more easily and quickly. They may be specifically designed to allow non-data scientists to incorporate AI features into their solutions quickly, or they may be intended to streamline deployment and management of custom machine learning or deep learning models. The platforms often offer stand-alone APIs for features related to text analytics, speech analytics, and visual recognition, including capabilities such as language translation, transcription, text classification, speech generation, video search, or facial recognition. They also usually ~include capabilities meant to streamline the building, training, tuning, and management of customized algorithms. They may incorporate frameworks, kits, libraries, and processing power into an integrated environment that speeds collaboration and deployment of machine learning and deep learning models. Furthermore, they may also offer operationalized solutions, such as chatbots, or solutions designed for specific business functions, such as sales or HR, or for a specific industry, such as manufacturing or  hospitality. Finally, the platforms may also be supported by complementing products and services that facilitate adoption of AI-infused solutions. These include hardware, such as cameras or edge devices, or professional services that help a business deploy their first AI project.

Build on solid foundations

As the ease at which AI can be incorporated into a multitude of enterprise workplace scenarios increases, as does the move towards cloud-based platforms and complex hybrid architectures. Often conceived in isolation by those “scary data-science folk” A.I solutions can be launched upon I.T operations with much fanfare, unfortunately a good portion of initial design concepts fail as the first real signs of operational stress materialise, basic considerations like good old-fashioned networking, application latency, cloud proximity, API integration and legacy systems interoperability present genuine challenges to the entire initiative.  Simple logic dictates that for every layer of automation added to a solution the question “what happens if?” becomes increasingly vital. Cloud is designed to fail and recover quickly, every major cloud brand knows this and designs their service accordingly. Successful adoption of A.I is not simply to focus on the endpoint, but to consider the whole picture, all cloud solutions regardless of use case demand a considered approach to design and operational integration. Technologies aside the cultural impact of A.I also needs careful handling, AI by its nature carries a sometimes, unwarranted negative vibe, fear of job-loss, personal work value to the business etc. The difference between a good deployment of A.I technology and bad will ultimately hinge on the buy-in of multiple stakeholders, change is painful at the best of times, but new technology and work practice automation needs careful planning against a work ethic that encourages inclusive behaviour as well as innovation.

A not-so artificial, but slightly intelligent cloud guy.