Generative artificial intelligence (AI) represents a significant technological shift that is poised to bring substantial changes to business spending in the coming decade and beyond. Despite the seemingly sudden impact of these changes, especially with the buzz surrounding conceptual AI in recent times, the transformation of organizational technology infrastructure must occur gradually and consistently.
As businesses prepare for resilience and functionality, the initial focus is on the infrastructure layer, which is currently experiencing significant investments in Nvidia and GPU aggregators. This trend indicates that the transition is already in progress. As development efforts advance, the attention will shift towards creating innovative experiences and products that will redefine each subsequent layer as adoption and investment progress.
There are early indications of a significant disruption at the application layer, signaling the beginning of a transformative phase.
Before the emergence of conceptual AI, enterprise applications had already started enhancing user interfaces by integrating interactive elements to engage users and streamline workflows. This evolution led to a shift from traditional “system of record” software like Salesforce and Workday to platforms such as Slack and Notion, now known as systems of commitment.
The anticipation of further advancements as conceptual AI shapes the next generation of application products is inevitable.
The latest enterprise tools have emphasized collaboration, offering features like online styling, identification attributes, version history, and metadata. These tools have not only enhanced user adoption but also facilitated seamless content sharing within and across organizations by leveraging popular consumer-centric components. Within these collaborative systems, the core data has maintained its intrinsic value, serving as the foundation for expanding the engagement layer and data volume.
Foreseeing significant progress as conceptual AI influences the future generation of application products is crucial. The initial entrants closely resemble ChatGPT integrators, developing lightweight tools directly on top of generative models to provide immediate albeit temporary value. While these generative AI products have shown rapid initial growth, they often face high churn rates due to limited workflows or lack of additional functionalities. These applications typically generate content or media that is single-use and rely on readily available generative models in the market.
The upcoming wave of relational AI applications, still in its early stages, will combine structured data from system-of-record applications with unstructured data from systems of collaboration.
Developers of these products have a better chance of establishing enduring businesses compared to first-wave competitors, provided they can dominate the layer above the systems of engagement and record-keeping. However, this task is challenging, especially with incumbents like Salesforce integrating conceptual AI to strengthen their core layers.
This sets the stage for the next wave, where participants will create their own “system of intelligence” layer. Startups will introduce innovative product offerings that leverage existing historical and engagement system capabilities to deliver value. Subsequently, they will develop processes that can eventually operate autonomously as robust business software after establishing a solid use case.
These products will generate new structured and unstructured data, using these datasets to enhance the product experience through generative models, essentially creating a new class of “super dataset.” This evolution does not necessarily involve replacing existing interactive or database layers.
The primary focus of these products will be on integrations that ingest, refine, and categorize data. For instance, simply extracting knowledge from current customer support tickets is insufficient to create a compelling user experience. A truly engaging product should cover bug tracking, product documentation, internal team communications, and various other features. It should be able to identify relevant data, categorize it, and analyze it to derive fresh insights. Moreover, it should incorporate a feedback loop to improve its performance through usage and training across multiple entities.
When a product meets all these criteria, replacing it becomes extremely challenging due to the invaluable, refined data it generates, a process that would be time-consuming to replicate with similar quality.
At this point, the product or concept not only contains intelligence but also includes elements like pyramid structures, labeling, and weights. By focusing on actions and decisions rather than just data production, these insights can be delivered within minutes rather than days. These authentic system-of-intelligence products, powered by relational AI, exhibit the following attributes:
- Ability to leverage previously structured and unstructured data and seamlessly integrate with organizational workflows.
- Proficiency in data classification and refinement through labels, weights, and sequencing.
- Establishment of data feedback loops within and across users to enhance the customer experience.
One critical question to ask clients is where a new product stands in relation to the other tools they use. Typically, the system-of-record product holds the highest importance, followed by the engagement method product, with additional tools at the base.
During financial constraints, the least critical product is usually the first to be removed. However, new system-of-intelligence products must offer lasting value to succeed. They will also face tough competition from established firms that will integrate relational AI-enabled intelligent features into their offerings. It will be crucial for the new generation of system intelligence to integrate their services with high-value workflows, collaboration, and the introduction of superior datasets to provide a compelling user experience.
The past year has seen a rapid evolution in the AI landscape, with businesses quickly adapting to the changing dynamics. Both proprietary custom models and open-source variants are proliferating at an unprecedented pace. In this rapidly evolving landscape, it is up to founders to create enduring system-of-intelligence products. When executed effectively, the impact on businesses can be profound.
