SANTA FE, NM - JULY 4, 2018: A man on a motorcycle wearing a Wenger Swiss Military Watch pauses at ... [+]
The debate around generative AI (gen-AI) continues to spiral, skew, twist, duck and dive. During this period of formative embryonic development and change, the industry has moved from its focus on guardrail concerns and open data access, onward to AI bias… and outwards to the scope for slimmed-down language models (large is not necessarily always best) created for industry-specific and task-specific use cases. That’s not to mention boardroom shenanigans and corporate sabre-rattling, but let’s move on. Perhaps its time to look at the implementation of gen-AI in temporal application use cases.
Known for its work in the FinTech space with data services aligned to the financial services sector and other verticals, KX is a specialist in vector and time-series data management. The company’s latest release is detailed as KDB.AI Server, a high-performance vector database for time-orientated generative AI and contextual search.
Deployable in a single container via Docker, KDB.AI Server offers setup options for various environments, including cloud, on-premises and hybrid systems. But what is time-orientated generative AI, where is it used and why does it matter?
What is time-orientated generative AI?
Put simply, time-orientated generative AI (with contextual search) is the creation of intelligence functions stemming from ingestion and analysis of time-series data sources i.e. those information pools with a specific time-stamped reference point on every piece of data as we might expect to find on weather records, energy measurements, financial stock indices, healthcare measurement systems or other similarly time-specific digital entities.
Because time dependency exists and pervades in the data topography handled and presented by time-orientated generative AI, the longer the period of time we work with and the more complex the relationships describes by the vectors used, the tougher the task is.
As data science commentator Fabiana Clemente writes here, “This time dependency introduces new levels of complexity to the process of synthetic data generation: keeping the trends and correlations across time is just as important as keeping the correlations between features or attributes (as we’re used to with tabular data). And the longer the temporal history, the harder it is to preserve those relationships.”
Where is KX KDB.AI used?
With all that time-orientated generative AI clarification (with a commensurate dose of contextual search also in the mix) under our belt then, if this is the tool, then where is it used in real world data and computing environments? KX says that KDB.AI's capabilities power versatile applications across a broad range of industry sectors including the previously highlighted financial services sector, where temporal and contextual search are used to augment trading strategies and reduce risk. This technology also works in gaming, plus also in e-commerce where real-time risk assessments and fraud detection matter a lot.
Other applications include healthcare and life sciences where the analysis of patient records can help lead to quicker diagnoses, personalized treatment plans and faster discovery of new drugs. In manufacturing and energy, KX points to the use of multi-faceted search for predictive maintenance, which helps to reduce machine downtime and improve operational efficiency. In aerospace & defence, we can see that analysis and correlation of operational data for intelligence helps improve command decision making. Also in government, this technology helps with ssarch and summarization of documents, video, audio and image files.
The ‘scale’ of the struggle
Generative AI promises to fundamentally transform productivity and drive competitive differentiation, yet as suggested by a recent report by Accenture, while 84% of global C-suite executives believe they must use AI to achieve their growth objectives, 76% report they struggle with how to scale. KX claims that KDB.AI Server solves this problem, giving enterprises the ability to drive their AI applications with data processing and search functionality that scales to meet the needs of the largest, most complex enterprises.
"The debut of KDB.AI Server Edition marks a transformative step in enterprise AI. It’s tailored for a future where data is a strategic powerhouse, enabling businesses to create custom AI solutions from their proprietary data to forge a distinct competitive edge,” said Ashok Reddy, CEO, KX. “Blending unparalleled data processing with agility and privacy, KDB.AI Server Edition isn’t just a new product, it’s a leap into the generative AI era, ensuring businesses not only adapt but also thrive and lead in the rapidly evolving AI landscape.”
Built to handle high-speed, time-oriented data and multi-modal data processing, KDB.AI handles both structured and unstructured enterprise data, enabling holistic search across all data assets with what the company insists is better accuracy. As a vector database, KDB.AI enables software developers to bring temporal and semantic context and relevancy to their AI-powered applications.
Further here, KDB.AI is optimized for Retrieval Augmented Generation (RAG) patterns which ensures that, rather than continuously training or fine-tuning Large Language Models (LLM), developers can bring data relevancy to their prompts delivering better accuracy, lower cost, and less need for GPUs. For completeness, let’s remind ourselves that - as detailed here - Retrieval Augmented Generation (RAG) can be described as an AI framework built to refine and improve Large Language Model (LLM) responses in terms of their consistency and quality by connecting the AI model to external sources of ratified knowledge data. In other words, RAG makes gen-AI more accurate.
Generative AI continues to develop and its alignment to time-series, time-stamped, time-specific data workloads makes logical sense in terms of its wider evolution and development. The deployment of this hugely impactful technology is widely agreed to be on its way to penetrating every application on our planet - it’s just a question of time.
