Lovelace this week emerged from stealth to launch a context engine builder, dubbed Elemental, that makes use of a proprietary YottaGraph to ensure the most relevant data pertaining to a task is exposed to an artificial intelligence (AI) agent.
That approach gives AI agents access to a context engine specifically designed to aggregate and index data in near real time versus always relying on large language models (LLMs) to reason across massive amounts of data in a way that, depending on the complexity of the task, can wind up consuming a massive amount of tokens, says company founder Andrew Moore.
At the core of that capability is a knowledge graph that has already aggregated and defined relationships between trillions of facts, also known as objects, that an AI agent can query on demand. Any time an AI agent is assigned a task, it can invoke Elemental to gain additional context within minutes in a way that serves to improve both reliability and safety, says Moore.
Underneath that platform is a database system that ensures that the millions of objects being most frequently accessed are made available in memory, adds Moore. Just as importantly, the YottaGraph provides a means to audit what data an AI agent accessed before it selected one potential outcome over another, he notes.
Ultimately, the goal is to provide more reasoning capabilities outside of the LLM, which in addition to improving accuracy, also serves to make organizations less dependent on any one provider of an LLM, says Moore.
While multiple vendors have already integrated knowledge graphs with AI models that are embedded within a specific platform, the scale at which Lovelace is making a general-purpose knowledge graph rivals in scale the search engine services provided by Google. It remains to be seen how YottaGraph stands up to what might one day soon be billions of AI agents regularly accessing data, but the alternative approaches being put forward by providers of LLM services are clearly already struggling to cost-effectively provide similar capabilities at scale.
The one thing that is certain is a massive amount of funding will need to be allocated to manage data in the AI era. A Futurum Group forecast projects the global data intelligence, analytics, and infrastructure (DIAI) market will grow from $409.3 billion in 2024 to $876.6 billion by 2029, achieving a strong 16.5% compound annual growth rate (CAGR). Exactly how that spending will be allocated remains to be seen, however, as new platforms for managing data emerge.
In the meantime, IT organizations will need to come to terms with the cost of AI. While it’s relatively simple to create an agentic workflow based on a narrow set of data, AI applications running in production environments will need to be able to access data at levels of unprecedented scale. Determining how best to cost effectively achieve that goal will ultimately prove to be the difference between success and failure when it comes time to evaluate the return on investment made in AI.
