Skip to main content

Identifying the MVPs in your AI deployment strategy

(Image credit: Image Credit: Geralt / Pixabay)

Artificial Intelligence (AI) has played a critical role in business transformation strategies. Once a figment of the human imagination, it has become crucial in improving processes, enhancing employee productivity, and revolutionising the customer experience. Based on this, decision-makers are turning to their IT departments and the professionals they need to deploy AI-integrated applications.

Many businesses leverage data scientists to execute AI-integrated initiatives because of their ability to interpret high volumes of data and change it into actionable insights. However, relying solely on them can be a crucial error in a business’ transformation journey. There are many obstacles that data scientists cannot face alone.

Decision-makers sometimes overlook the other integral team members when developing their AI-strategy. The MVPs in AI implementation processes are the 4 D’s: data scientists to leverage data and create models, designers to work on UI and UX, developers to build the right software or application that applies the models to business processes and DevOps to manage and update the infrastructure, integrations, deployments and model management.

To avoid overlooking the right resources during the transformation journey, a phased approach should be implemented. It will ensure that all bases are covered, from initial development stages to AB testing and finally, integration throughout business processes. Let’s take a look at this in more detail.

Phase 1: Data hoarding 101

When embarking on an AI deployment journey, it is crucial to recognise that AI would not exist without the capability to analyse and draw meaningful insights from large volumes of data. Data collection is, therefore, a critical step in this process. The next step is to acquire the infrastructure and make sure machinery is enabled to support this. The DevOps team will have a hand in ensuring a business’s infrastructure is optimised to collect and store data.

Data scientists take on another element of the strategy, by figuring out the best way to leverage the data. Typically, they will comb through the data, looking for insight at a deeper level, building predictive models and, ultimately, giving recommendations that could impact the business. All this takes place within a research environment that prepares the ground for the actual implementation of the strategy.

While this may seem straightforward, things become more complicated when dealing with large masses of data. Let’s use the Industrial Internet of Things (IIoT) as an example.  Machinery can produce copious amounts of data per millisecond, which is why the creation of Machine Learning (ML) integrated analytical models that can replicate analysis on a much larger scale is vital. When leveraging both AI and ML, it is important to implement these models for these functions to work.

Phase 2: Testing 1, 2, 3

Now that the models are in place, what happens next? This is when it’s time to start thinking about how the other roles fit in to execute an AI strategy. After being established and deployed, the models have to be properly managed and updated within a dynamically changing scenario, with the aim to roll them out and be productionised. This is where the developer team comes into play.

Testing the models during the production process is where software engineers shine. The developer team can leverage valuable insight from testing which can be applied to the business processes. Here, it is important for organisations to understand that this is a software engineering problem, and not just a data science problem as is quite commonly thought. This step is another crucial component that data scientists could not do without major pitfalls. The production process requires a considerable amount of organisation and ongoing optimisation on data analysis. The developer team can then explore how data analytics models can be integrated and updated without disruption. From this, predictions and insights stemming from the model testing can be rolled into business processes to optimise them and bring real value.

It is imperative for developer teams to be involved in order to create the right software and/or applications that can be applied to the business as well as the end-user. These solutions are a form of aggregated knowledge, but can only be valuable when streamlined and integrated into business operations. However, this can only go so far without the help of the design team.

Phase 3: Art & design

The designer team’s role is to ensure that the software and applications developed have an aesthetically appealing and user-friendly user interface (UI). This step in the AI-deployment strategy could potentially make or break the initiative, and should not be taken lightly.

Phase 4: Monitoring the digital infrastructure

Deploying AI is not a one-off project with clear date of completion. Far from this, it is a comprehensive strategy that needs to be managed, maintained, monitored and fine-tuned on an ongoing basis to ensure the data collected and analysed reflect real-time information. The DevOps team is responsible for constantly monitoring and updating the infrastructure, while maintaining a line of communication across all departments to ensure all the key stakeholders are kept in the loop. This is another vital piece in the AI-deployment puzzle, which can ensure the seamless functioning of all new integrated models. Traditionally, the DevOps team will continuously monitor the applications in the production environment and prepare for typical monitoring occurrences. With full visibility of the infrastructure, this department can optimise all areas of the business leveraging the insight taken from the data models and detect anomalies that could harm mission critical applications before they even occur, protecting the entire business from errors and/or downtime.

 Collaboration is key

The latest innovations such as AI and ML can seemingly revolutionise a business and enhance its offering and processes exponentially. However, in order to do this successfully, businesses must invest not only monetarily, but consider other factors to execute technology driven initiatives properly. Lack of knowledge and understanding can be the crux that derails AI-deployment, and will need to be addressed if businesses want to move forward with their transformation journey.

Having a holistic view of a business’ AI strategy is imperative for evaluating roles and responsibilities.

Ruban Phukan, VP Product, Cognitive - First, Progress (opens in new tab)
Image Credit: Geralt / Pixabay

Ruban Phukan is the VP Product, Cognitive- First at Progress where he leads product and the data science for the flagship Cognitive Predictive Maintenance product which solves the complex business problems of minimising asset-failures / unplanned-downtimes.