Instead of just another decimal-point upgrade to a massive language model, the latest AI move from publisher Wiley and AWS is a different breed: It’s a targeted agent from publisher Wiley and AWS: an AWS-hosted agent researchers can fork on GitHub and plug into their own pipelines. Wiley specifically mentioned its potential to cut research tasks like extracting detailed methods or results from multiple papers “from days to minutes.”
Touting the tool as the “first of its kind from a publisher on AWS,” the AI agent will make its debut at the latest AWS Life Sciences Symposium slated for May 6 in New York City.
Wiley is making the literature search agent available as part of an AWS-assembled open source toolkit for healthcare and life sciences agents.
The code lives in AWS’s new health-and-life-sciences agent toolkit on GitHub, released under the permissive MIT-0 license and already attracting forks from biotech scientists on GitHub. This toolkit, which AWS detailed in December 2024, is designed to orchestrate multiple specialized AI agents for complex workflows like cancer biomarker discovery, combining database analysis, statistical modeling, and literature search. Wiley’s contribution adds a dedicated agent for querying its own trusted content within this collaborative framework.
The collaboration integrates “Wiley’s authoritative full-text content into AWS’s technology built on Bedrock Agents” to “bring critical scientific insights directly into researchers’ existing workflows,” said Josh Jarrett, SVP and GM of AI Growth for Wiley.
Explaining the motivation behind the collaboration, Dan Sheeran, General Manager, Healthcare and Life Sciences at AWS, put the challenge thusly:
The cure for cancer isn’t going to come from an abstract, but will be derived from researchers interrogating and synthesizing internal and external data.
- Multi-agent collaboration. Left column: Research scientists ↔ Amazon Bedrock supervisor agent, which runs a “chain-of-thought reasoning” sequence (Step 1…Step n). Right dashed box: five sub-agents, including biomarker database analyst, statistician, clinical evidence researcher, clinical trial analyst, medical imaging expert, each wired to domain tools (Amazon Redshift, custom container/code interpreter, PubMed KB, ClinicalTrials.gov, Amazon SageMaker job). Image from AWS.
