How platform technology accelerates your AI projects

Many organizations have been experimenting with Artificial Intelligence (AI) and machine learning to make processes and services more efficient. However, most of them are also struggling to move beyond a proof of concept and integrate models into their business operations. Fortunately, there is a solution. With the right platform, you generate faster and better value from your AI projects.

Artificial intelligence may sound like a buzzword nowadays, but from a technological perspective it is nothing new. Statistics, analytics, big data, … These are all different names that refer to a climate in which organizations are gradually becoming more data-driven. Recently, digital transformation has accelerated and the current pandemic has made it clear to companies that they need data and AI to make their business more resilient. They understand that automation can help them become more efficient and improve the customer experience, for example with personalized communication.

Many companies have experimented with analytical models, but they often remain stuck at a proof of concept. In fact, they usually experience three major obstacles.

The 80/20 rule

First, people often think that data scientists are constantly developing and implementing models. In reality, however, they only spend 20% of their time on building models or dashboards. The other 80% is spent on preparing and cleansing the data before it can be used for analytics. This is very time-consuming work and a huge barrier to even getting started with analytics. Organizations can bypass this obstacle with platform technology. Since most modern businesses want to become leaner and simplify their IT landscape, they embrace platforms to innovate and generate value faster.

When you add AI use cases to industry solutions and standardize the AI capabilities for an industry data model, you do not have to prepare data. For example, a Consumer Goods company has field sales teams that need to visit bars, restaurants or stores to see if their products are properly displayed or to check if promotions have been activated. With a powerful platform, they can streamline these field sales activities and immediately see which places they should visit to increase sales or predict which promotions will be most effective. If you have to first prepare data and integrate multiple legacy systems, you lose a lot of time.

Operationalizing a model

Data preparation is a major bottleneck, but this is just the first step of the AI lifecycle. An analytical model is a means to an end and can only generate value if it is integrated in operational business processes. For example, you may have built a nice model that can predict why certain customers are leaving, but now you have to make sure that salespeople or staff in a contact center can take advantage of this model. If they know why customers are at risk of leaving, they can offer special discounts or proactive actions to increase their satisfaction.

In many cases, a model never makes it beyond a proof of concept (POC) because IT is unable to operationalize it. And even if they succeed, a model loses predictive power over time and requires a lot of maintenance and monitoring to adapt to new circumstances. With the right platform solution, you can immediately operationalize analytics and integrate it in workflows and business rules. It automatically offers the front-end engagement layer that salespeople, contact center agents, marketeers, partners and suppliers use every day.

Data scientists

A third and final obstacle: data scientists are very scarce on the job market. Only a minority of organizations with interesting use cases is able to attract and retain data scientists. Platform technology allows companies to use analytics without interference of a data scientist. It generates straight-forward predictions and recommendations that can be used by business analysts in your organization. By making AI accessible, companies without data scientists can start implementing AI use cases. And for companies with data science departments, it can free up some time from data scientists by offloading work to business analysts within the organization.

The Salesforce platform enables companies to accelerate the implementation of AI in sales processes. It comes with an industry data model, industry processes and embedded AI capabilities, eliminating the need to prepare data. Together with Einstein – a smart assistant giving employees fast access to new insights, next best actions, and recommendations – models are efficiently operationalized without the need for a data scientist. The Trailhead platform also provides several learning trails on Einstein and how to use this technology.

The platform even adds extra context around predictions and explains how certain results are obtained. An in-built alert mechanism also points out potential biases in model building, enabling business analysts to build bias-free, accurate and understandable AI models. Explainable and Ethical AI are two essential concepts, but this is something I would like to come back to in another blog.

Are you interested in becoming more data-driven? Then get in touch with me to learn more.