Five key steps to achieving real business benefits from GenAI Adoption
By George Dragatsis (pictured), ANZ Chief Technology Officer at Hitachi Vantara
The pace of adoption of Generative AI (GenAI) by organisations of all sizes continues to grow. Increasing numbers are coming to understand the business benefits the technology can offer and are keen to put it to work.
It is this growth in usage of GenAI that’s poised to change the face of how computing technology is utilised. It is going to require a change in strategies and the establishment of new supporting platforms.
There are five key steps enterprises can take to accelerate their GenAI adoption to drive the collective Artificial Intelligence (AI) future forward. These steps are:
- Know your goal:
Enterprises generally start small with new technology initiatives and look for everything that’s required to steadily scale and achieve ROI. For GenAI projects, whether in the financial services sector, manufacturing, healthcare, or transportation, the winning formula will include a tailored solution suite designed for industrials and businesses committed to AI emergence. Unsurprisingly, such a suite must empower organisations to automate processes, accelerate time-to-insights, and unlock innovation.However, the bespoke piece is the customisation required within a particular industry and built on top of versatile foundational models and technology. Wherever an enterprise is along its AI journey, it will require a combination of platform, infrastructure, software, and sometimes support services to go up the stack as necessary. Most organisations don’t have the in-house expertise or ability to handle all of this by themselves. Wrapping everything together into a single solution built “outcome-facing” – from UX all the way down into the core infrastructure – is the ideal.
- Combine core enterprise standards:
According to recent research, approximately 9 out of 10 AI projects are failing, sometimes due to wrong technology choices, change management issues, unforeseen risks, uncertain ROI, and/or dried up funding. Therefore, democratised access to all of the attendant technologies simply isn’t enough to get or keep the AI engine on the racetrack. Knowing how to properly use those technologies is at a premium.Savvy domain experts need outstanding time-to-value, but don’t have AI/ML expertise. Enterprises are seeking a new kind of specific technical guidance on accessing enterprise data for fine-tuning, RAG, or adaptive pretraining. They also need guidance on enterprise-specific models to perform certain tasks, virtualized access to distributed data across a hybrid cloud ecosystem, and achieving end-to-end visibility into their overall system.
- Support “data anywhere” and “AI everywhere”:
Getting the data part of the equation right is especially critical for accelerating GenAI adoption — and it’s the most challenging part. Consider the whiplash enterprises have been through. They started off in the datacentre, then started moving aggressively to the cloud. Then, with the cloud’s promise unrealised amid high expenses and resource waste, they retained (and in many cases returned to) hybrid deployments. Their data has been on this journey with them, so grasping where their data actually lives, how it’s best managed, and how they can best utilise it to effectively answer their AI questions is key. - Use a high performance, load-optimized, AI-ready appliance:
Solutions built on AI require a tremendous amount of horsepower. The fastest and most effective platforms to drive the enterprise AI journey will support systems like NVIDIA DGX and other supercomputers that can process trillion-parameter models for massively scaled GenAI training and inference workloads. They’ll require very powerful, super-fast file systems behind them and economically viable object storage.
Enterprises will be able to solve their AI compute and storage challenges by bringing these components together in an appliance with high performance and load optimisation for cost efficiency. AI-ready infrastructure engineered in this way embraces outcome-led design, offers accelerated integration based on a pre-built foundation, and provides bespoke customisation and delivery. That’s a package prepared to solve specific industry challenges and future-proofed for new ones that arise.
- Choose an AI partner-provider ecosystem:
In joining the AI race, few enterprises, if any, are going it alone. Enterprises exist in their vertical markets and each wants to accelerate their transformation with GenAI solutions that make sense for their space. They want an infrastructure and data partner with domain expertise in these areas that understands the industry-specific questions they’re trying to answer.Important too in the partner ecosystem is that an enterprise’s AI partner-provider is “eating their own dogfood”. That is, they’re building out their own AI solutions, like copilots and companions, to improve their internal processes. Doing that moves technology from a reactive position to a proactive position in what it actually means to adopt AI.
It’s clear that AI is going to deliver significant benefits to organisations that strategically choose and deploy the right mix of underlying technologies in the months and years ahead. Taking time to carefully plan projects now will significantly improve the likelihood of future success.
