Amway’s goal is to deliver the maximum impact it can to support people’s health and wellbeing. The company holds over 750 patents (granted and pending), operates 75 research and development labs around the world, and employs 800+ scientists, engineers and technical professionals working across a vast range of disciplines. In order to allocate R&D spending effectively, decision-makers require both a broad and deep understanding of cutting-edge science.

Jesse Leverett, Fellow of Open Innovation and Chair of Amway’s external Scientific Advisory Board, described how before HASH the company’s scientific literature review process was manual and time-consuming.

At a meeting of Amway’s Scientific Advisory Board in mid-2024, the board recommended that a larger-scale structured analysis of scientific publications be conducted, enabling the impacts of exposomal factors (e.g. smoking, or air pollution) on human healthspan and intrinsic capacities (e.g. motor function, or cognition) to be systematically traced through to their specific effects on human cells, tissues, organs, systems and processes. With this information, Amway hoped to identify commonly-overlooked relationships with a strong scientific evidence base, suggestive of their high potential value as R&D targets, and identify novel pathways of effect that might be used to prioritize research into new products.

However, comprehensively surveying academic literature, and identifying the millions of potential relationships described within, was a task beyond any one human or team.

Amway’s use of HASH enabled them to automate the analysis of over 10,000 scientific papers, assembling a strongly-typed, checked knowledge graph in days, rather than months, with a throughput equivalent to hundreds of scientists working in parallel.

Darius Machado, Director of Nutrition Portfolio/Labs/R&D Business Intelligence/PXI, said:

Beyond the initial AI research required to populate Amway’s web with data, insights from newly-published papers relevant to the company’s scientific domain can now be continuously integrated in the background, on autopilot, ensuring that the graph remains useful, relevant, and up-to-date. Without HASH, it is estimated that the cost of producing an equivalent initial knowledge graph could have cost upwards of $1.25m, with ongoing costs and demands on labor beyond this.

Within days of engaging HASH, Amway had a knowledge graph consisting of more than 100,000 entities, each with its own set of corresponding structured data attributes. However, rather than overwhelm users, HASH’s in-built data analysis algorithms enabled Amway’s team to quickly explore the graph data themselves, clustering entities into related groups, and identifying pathways between them.

Using HASH to view data in both conventional tabular spreadsheet, as well as “non-linear” graph form, Amway researchers are able to quickly sort and filter results, and easily pull up detailed views of individual entities, extracting actionable insights in seconds.

HASH lets users see the provenance of individual datapoints, and quickly jump to relevant sources, all without leaving the platform or needing to log into external academic databases.