The Covid-19 pandemic has advanced digital transformation by three to four years across different areas of business, as leaders have been forced to invest and upgrade digitally.
As leaders look to build on their digital platforms, says Abhinav Joshi, director of AI strategy and GTM at Red Hat, artificial intelligence (AI) emerges as a great enabler for businesses looking to generate efficiencies, improve customer experience, increase revenue and save costs.
We are seeing artificial intelligence be applied across multiple industries, such as financial services and insurance, telecommunications, healthcare and automotive. The greatest impact comes through AI strategies that prioritise people and processes as well as technologies. Here are five key considerations to help businesses to get the most value from their AI investment.
1. Prioritise the quality, not quantity, of use cases
Leaders should ask themselves which specific operations would gain the most from being digitised or enhanced using AI, then proceed to select one or two projects to target first. For example, a healthcare provider might benefit from building a data-driven diagnostics solution for a medical condition, while a banking service might prioritise a fraud detection solution for credit card transactions.
These initiatives should be properly funded, rather than having data teams simply experiment with capabilities. For a greater chance of success, all stakeholders should be invested in the outcome; having a funded initiative will help to encourage this, as will sharing the project’s business goals with the rest of the company.
2. Dedicate capacity and talent
Once a business problem has been identified and project aims have been outlined, leaders must turn their attention to their team. This team should be capable of leveraging all the tools and technologies needed to carry out the project and will likely be made up of a selection of business analysts, data engineers, data scientists, machine learning (ML) engineers, developers and relevant IT operations specialists including perhaps AI software ecosystem partners.
It’s important to weigh up internal upskilling programs against external contracting, as while the project may be short term, its use will most likely be long term. Process changes in operations and governance must also be considered at the outset.
3. Set up for scale
A hybrid cloud strategy helps ensure consistency and flexibility when scaling and managing workloads across data centre, edge and public cloud environments. Choosing a Kubernetes-powered hybrid cloud platform with hardware assisted acceleration (such as Red Hat OpenShift) can give data scientists, application developers and ML engineers rapid, self-service access to resources when building the AI solution.
This is also the time to gather and prepare the right data for the project, and establish how it will flow through your architecture. Businesses collect huge amounts of data, so cleaned, relevant data will enable effective creation and testing of new AI models. Alongside gathering suitable data, leaders must establish adequate compute storage and hardware accelerators.
4. Automation and access
It is essential to consider how to automate the entire ML lifecycle for the future and who will have access to tools and key workflows. Appointing an enterprise architect or similar to design the architecture is important, and ensuring relevant teams have access to appropriate MLOps (DevOps for Machine Learning) tools in a self-service way will make the journey smoother.
Turkcell sets a shining example in this space, it built its AI services architecture and application hub on a hybrid cloud platform running across its business to enable a consistent, self-service experience for data scientists and application developers wherever they are in the business.
By making its AI capabilities accessible from anywhere in the organisation, Turkcell has been able to empower data scientists across teams to benefit from AI capabilities and spur innovation in diverse areas, from digital customer onboarding, to charity awareness campaigns, music streaming services and fraud detection.
5. Measure your success
At the outset of developing an AI strategy, tangible success metrics and criteria should be established to ensure the project remains focused throughout. Teams must make sure the recommendations produced by the machine are accurate and explainable. This serves to legitimise its outcomes and provides valuable proof points. Having the ability to show the ROI of the initiative can help encourage its wider rollout.
Making theory a reality
AI has great potential to transform industries, societies and lives. Business leaders looking to embrace the emerging opportunities as the technology evolves would do well to adopt a ‘pilot, prove, expand’ strategy that brings the entire company with them on the journey.
The author is Abhinav Joshi, director of AI strategy and GTM at Red Hat.
About the author
Abhinav Joshi, director of AI Strategy and GTM in the OpenShift business unit at Red Hat. He is focused on cloud-native workloads on OpenShift, including AI/ML, data analytics and databases. Abhinav has over 19 years of industry experience focusing on hybrid cloud, AI/ML, data analytics, digital workspace, and software defined data centre products and solutions.
Throughout his career, Abhinav has held strategic roles in product management, product marketing, sales, and consulting services at Red Hat, VMware, Cisco, NetApp, and more. He holds an MS in Systems Engineering from the University of Maryland College Park, Strategic Management Graduate Certificate from Harvard University, and a B.Tech in Chemical Engineering from Nagpur University, India.
Follow us and Comment on Twitter @TheEE_io