AI in the enterprise: 4 strategies to make your big push pay off

Despite the recession, companies are still committed to investing in AI. They aren't just buying the technology, either; they're hiring experts, identifying the right places to deploy it, and initiating proofs of concept.

Despite the flurry of activity around artificial intelligence, however, a study from Capgemini found that four out of five organizations fail to successfully scale their AI programs (PDF document) beyond the pilot and initial production stages.

AI programs can easily pay for themselves within their first few months on the job, but programs must be implemented in a thoughtful and strategic way throughout the entire enterprise (including areas such as AI in testing) to reap these benefits. AI does its best work when it is not confined to a silo. 

If AI isn’t given the proper runway to show what it can do, initial lackluster results can lead to decreased funding, which can jeopardize the program's long-term returns.

The potential for AI to transform businesses across all industries is obvious, but there are a few common roadblocks preventing it from reaching its potential. Here are four strategies you can execute to ensure that your company's investments in AI reach their potential.

1. Leadership must be invested in AI and its success

Building AI models is only the beginning. Deploying those AI models properly requires additional resources, including the right staff and structure within the company to keep things whirring.

Leadership is critical to this effort, since the process and investment in AI needs executive sponsorship to stay on track and receive the attention necessary to thrive. To create a compelling case for decision makers and get the support you'll need to continue innovating and scaling those programs within your company, your AI teams must demonstrate the value their programs bring to the organization in tangible, bottom-line, relevant ways.

2. An AI team must be more than data scientists

While data scientists are critical to the successful implementation of AI, team members with other skill sets are also vital to the effort. Your team will need data engineers and machine learning (ML) engineers to build the pipeline and get models in production, respectively.

To satisfy the need to prove the ROI for executive leadership, you also need business analysts on the team to capture the insights from the data and produce the numbers that show positive business impact. Ensure the AI team has a diverse set of skill sets—such as in statistics, econometrics, back-end programming, ML, and data engineering. This is paramount for AI efforts to benefit the organization as a whole.

3. Choose the right technology to support AI

To get AI models up and running, companies need to choose the right technology and architecture to support them. Make the right investments to set up the right technology stack. 

ML-based ops tools are helping companies with the entire lifecycle of ML models, from managing the models to deploying the models in production and monitoring how models are performing in real time. With the right supporting technologies in place, your teams can execute the pipelines for continuous integration (CI) and deploy the models at scale.

4. Build an AI center of excellence

Data scientists and engineers are often embedded within specific IT or business functions throughout companies. While this deployment is practical in theory, it can also create silos. Not allowing the data scientists to see across the company and connect with their colleagues can leave them biased toward their own model. 

In this scenario, data scientists are developing their own ML models using their own, preferred tech stack. This does not allow for collaboration, model-sharing, or documentation of learnings for future model building.

Creating an AI-centric operating model can remedy these issues, commonly known at Capgemini as the AI center of excellence. This is responsible for the end-to-end lifecycle of AI projects, seeing them from concept to completion—or in AI terms, from pilot to production to scale. Most enterprises lack this kind of central governing body that helps ensure that projects are completed and deployed effectively.

Put the effort in and it will pay off

When used properly, AI can benefit companies both from an efficiency and a profitability standpoint. It is not as simple as taking AI out of the proverbial box and turning it on, however.

You must put in the effort to get leadership on board and invested in its success, choose the right team across disciplines to manage it, deploy the correct supplemental technologies to support it, and create a governing center within the company to see it through. If you do, the rewards AI can provide will significantly outweigh the initial investment it requires.