We stand on the verge of a generative AI (GenAI) revolution. Some 98% of organizations have specific GenAI goals for 2024 — accounting for nearly a third of digital modernization spend last year and in 2024, according to new research from Couchbase. According to our survey, CIOs are excited by the prospect of productivity enhancements, rapid prototyping and customer experience (CX) improvements, among many other touted benefits. But can existing IT infrastructure support large-scale AI projects?
The answer for many CIOs is “no.” To usher in a new age of GenAI-powered adaptive applications, organizations must first modernize their data management strategies to gain control over the high-speed data analytics and processing that AI increasingly demands.
A Limit to Growth
Our research indicates the average investment in digital modernization per organization was $28 million in 2023, set to grow 27% to exceed $35 million this year. Yet technology, resource and organizational buy-in remain key barriers. We calculate that, on average, organizations waste $4 million annually on failed, scaled back or delayed projects. Nearly two-thirds (63%) have suffered delays of longer than three months due to IT modernization issues.
When it comes to GenAI, even just a few months could be the difference between long-term business success and failure. Enterprises need to increase productivity by more than a third each year just to remain competitive. They can’t afford projects to fail or be delayed.
The challenge is that most organizations are not set up to support the kind of next-gen adaptive AI apps that will transform user experience through hyper-personalization and real-time updates. They don’t have the necessary security and privacy guardrails in place. They can’t deliver the low latency needed for rapid data access, sharing and usage. And they don’t have multipurpose databases that help to reduce GenAI hallucinations by creating a single pool of trustworthy data to interact with external models.
Putting the Pieces in Place
CIOs must consider change not just at a technical level, but also at a cultural one if they want to drive GenAI success. This will require them to set realistic goals and expectations for what the technology can do to get senior leaders on board and give projects the best chance of success.
Speed is also critical. Sharing and accessing data must happen rapidly for peak performance. If not, apps will serve up outdated information, and there’ll be a bigger risk of hallucinations. Security and privacy are also crucial: CIOs need to prevent inadvertent exposure of sensitive intellectual property. Finally, it’s vital not to forget the end users of the technology. Employees must be trained to ensure optimal and safe use of GenAI.
Maintaining and improving GenAI capability without reducing investment in other areas will be a challenge. But it can be done. Focusing on data architecture is a good place to start. It’s notable that over half (54%) of enterprises admit that they currently don’t have all the elements in place to ensure an all-encompassing GenAI-ready data strategy. There are multiple aspects to consider.
Organizations first need control over where the data is stored, who has access and how it is used so that it can’t be accessed or used inappropriately. They will also need specialized tools and procedures in place to prevent proprietary data and customer information from being exposed outside the organization. Developers should be given clear and detailed best practice advice to use data safely and effectively.
Next, consider the data architecture itself. As discussed, it’s vital that any system can support real-time GenAI applications by ensuring data can be accessed, shared and used with minimal latency. A high-performance database that can manage unstructured data at high speed will prevent GenAI from being limited in how it queries data. This will also support near real-time data analytics — another vital requirement for GenAI to provide accurate answers to its users. Only 18% of enterprises have a vector database that can store, manage and index vector data efficiently, but this would also help improve GenAI performance.
However, it’s worth remembering that GenAI often requires different levels of data processing. So IT infrastructure must be able to scale to meet immediate demands without incurring unnecessary spending. Finally, consider the challenge of hallucinations. Organizations should consolidate their database architecture to prevent AI applications from accessing — and becoming confused by — multiple versions of data. Less than a third (31%) of enterprises have made this investment.
It’s Time to Create
With these pieces in place, organizations can think seriously about creating GenAI-powered adaptive applications. These are apps that perform a single task but use AI to intelligently, dynamically and automatically adapt to changing circumstances and their users’ particular preferences. A booking app may regularly update based on real-time travel information, events and a user’s history to suggest journeys and personalized deals, for example.
We estimate that half (46%) of businesses will lose customers and 36% will hemorrhage staff to rivals if the applications they deliver no longer meet expectations. With these expectations rising all the time, businesses can’t afford to stand still. They must deliver the kind of hyper-personalized and contextualized experiences these users demand, or risk an existential threat to their business.
It Starts With Data
Getting there won’t be easy. But there are database technologies out there right now to support these ambitions. These multipurpose platforms offer control over data storage and access, can manage structured and unstructured data at high speed, scale on demand, and support technologies like vector-based search and real-time analytics. They also support edge computing for high-speed data sharing and access and enhanced security.
Rather than the single-function databases of old, they offer everything an organization needs to support their push for adaptive applications. The future is just around the corner.
Learn more about how Couchbase vector search at the edge and real-time analytics with Couchbase columnar can help organizations develop a new class of AI-powered adaptive applications that engage customers in a hyper-personalized, contextualized way.
The post How Adaptive Applications Unlock Innovation in a New AI Age appeared first on The New Stack.
Adaptive apps use AI to intelligently, dynamically and automatically adapt to changing circumstances and users’ preferences.