Not too far into the future, the apps on our smartphones will be AI-powered for hyper-personalized experiences.
Mobile apps have transformed the way we live, work, and interact. But it doesn’t end there. The explosive interest in generative AI is opening a new world of functionality for user-centric applications that can “adapt” their behavior and features based on user preference, situation and/or context.
But are we truly ready for AI-powered mobile apps?
DigiconAsia talked to Genie Yuan, Regional Vice President, APAC, Couchbase, about being ready for AI-powered adaptive applications, how they are transforming businesses in the region, and their impact on data security and privacy.
How have mobile apps transformed businesses, especially in Asia Pacific?
Genie Yuan: The biggest impact has been in how businesses interact with customers, and even more so now that artificial intelligence (AI) features more prominently in app development. But even before the recent advances in AI, apps were pushing the envelope on personalization. Apps were providing the foundation to launch ad campaigns based on users’ locations and preferences.
Apps have also been instrumental in raising scalability and agility. Brands have been able to tap into them to meet customers where they are. Now with AI, we are noticing that brands at the front of the pack are utilizing apps to deepen connection with users, taking in their feedback, and giving them a voice to shape app interface. And, of course, this depth of connection demonstrates that these brands care about their customer’s feedback, which drives sustainable business growth.
How are adaptive applications different from existing mobile apps? What are some examples?
Yuan: Adaptive applications are equipped with features and interfaces that constantly change in order to meet user preferences or data input. By tailoring capabilities according to real-time needs, adaptive applications can create a hyper-personalized and responsive customer experience.
An example of this can be seen with streaming services. Most streaming services can suggest content based on users’ viewing history and preferences. However, those with adaptive capabilities can differentiate themselves by automatically scheduling TV watch sessions, pausing shows while viewers are away, or teasing significant events on another channel.
e-commerce is another example. Typically, e-commerce platforms use purchasing and browsing data to recommend items that shoppers are most likely to purchase. However, adaptive e-commerce applications can also help sellers create customized descriptions that align products with customers’ needs. This feature is incredibly useful, especially when buyers are making a purchasing decision.
How ready are businesses to adapt AI into their mobile apps? Can they trust the data that employees are feeding to GenAI?
Yuan: Barring outliers that are ahead of the curve, most businesses are lacking in AI-readiness. These laggards are not prepared to minimize the security risks involved with feeding data into large language models, for instance.
There are also concerns about data complexity when building prompts. A disparate environment comprising purpose-built databases, as well as data manipulation and cataloguing tools hamper personalization. Many expect to just deploy AI and hyper-personalization will follow, but multiple databases may mix unrelated or contradictory information that could lead to hallucinations.
Organizations will also need to prepare for complexity shifting away from data architectures to how AI interacts with public models, especially as the latter evolves, diversifies, and becomes contextually specific.
How can organizations meet the data security and privacy requirements when building AI-powered apps?
Yuan: In terms of data, organizations can encourage data consolidation by integrating JSON-based databases that can instantly adapt to new data types and models. This empowers organizations to reduce the number of data sources they have. With this feature, organizations can gain accurate insights and shorten app building and scaling times.
Besides that, organizations should integrate vector search, which determines what information LLMs are allowed to share and which areas they should retrieve data from. Not only will this feature enable users to gain relevant results quickly, but it can also protect proprietary data from being exposed. This, in turn, can help organizations reinforce user privacy.
Lastly, safeguards need to be prioritized in configuration. Role-based access controls (RBAC), which restrict data authorized users and applications, goes a long way here. Besides that, having intrusion detection systems in place enables security teams to stay on top of unauthorized access attempts, identity changes, and security configurations’ performance.
