AI Firm Revolutionizes Research Access for Industries Worldwide

A new wave of artificial intelligence tools is transforming how research is accessed across various industries. The AI firm, Custom D, is developing innovative assistants that help organizations tap into vast amounts of information previously hidden within extensive reports and studies. Among its notable products is the AI assistant known as Caitlyn, designed for the Foundation for Arable Research (FAR), and Bella, created for Beef + Lamb New Zealand (BLNZ).

These tools are aimed at making research more accessible to users, a need that has become increasingly apparent as research efforts often remain underutilized. Julie Ryan, chief executive of Custom D, noted that the organization has been working diligently over the past 18 months to build a robust AI platform to meet this demand.

Maximizing Research Potential

Custom D was established as a spin-off from Digital Fusion, co-founded by Craig Saunders alongside Ryan, Sam Sehnert, and Matthew Rhodes. The company plans to scale its AI offerings this year following the successful development of pilot programs. During a recent visit to the United States, Ryan highlighted a staggering statistic: the U.S. alone conducts approximately $1 trillion in research each year, yet only 14% of that research is effectively utilized.

This underutilization, she pointed out, poses significant challenges, especially when the research has the potential to save lives, mitigate environmental impact, and enhance food production. As AI technology evolves rapidly, tools like ChatGPT have become commonplace; however, creating a system that maintains integrity and security while handling extensive datasets presents a unique challenge.

Ryan explained that Custom D recognized the opportunity when large language models like ChatGPT and others emerged. The company began collaborating with FAR in 2024 to deliver three decades of research data, enabling farmers to improve crop yields.

Innovative Solutions for Complex Data

Ryan emphasized that the research data is often technical and lengthy, making it difficult to navigate. The use of generative language models, combined with retrieval-augmented generation, allows for the summarization of vast quantities of information into a more digestible format. The Ask Farai assistant, now in production, is designed with security as a priority, protecting sensitive information while providing users with reliable answers and source links.

One of the standout features of Caitlyn is its capacity to operate within a client’s private cloud environment, which allows organizations to control their own knowledge base. This approach ensures that users can manage their data without relying on external web sources. The interest in Caitlyn surged following the launch of BLNZ’s tool in September, further solidifying the demand for AI-driven solutions in the agricultural research sector.

As Custom D continues to expand its client base, it has engaged with several research organizations across New Zealand and is in discussions with one of Australia’s largest contract research organizations (CROs) to initiate a pilot project. The company has also been in talks with agricultural leaders during its U.S. visit, identifying opportunities for collaboration with research institutions in North Carolina and Boston.

Beyond its current capabilities, Caitlyn offers multilingual support for non-English speaking countries. Custom D’s achievements were recognized at the AWS re:Invent conference in Las Vegas, where it won both the innovation partner of the year and social impact partner of the year awards.

With a dedicated team of 25, including developers and data scientists, Custom D is actively working on approximately 10 pilot projects alongside another 10 in production for clients such as Deer Industry NZ, NZ Avocado, and Forest Growers Research. Ryan remarked that entities engaged in extensive research, including input providers and public agencies, would greatly benefit from integrating Caitlyn into their systems.

The development of this AI platform has involved over 15,000 engineering hours, underscoring the substantial investment in making research more accessible. Looking ahead, Ryan expressed enthusiasm about the potential for future enhancements, including integrating with additional systems to streamline workflows. “What can be done on top of that — the world is your oyster,” she stated, highlighting the limitless possibilities that lie ahead for AI-driven research solutions.