How Manifold AI Built a Drug Discovery MVP on AWS

In life sciences, the time between a research question and a usable answer is often measured in weeks or months. For patients waiting on new treatments, that time has a real, tangible impact. Critical workflows depend on large teams to execute manual, repetitive steps, such as gathering data, running analyses, reconciling results, and extrapolating conclusions. At institutions like the UVA Comprehensive Cancer Center, a single workflow sometimes required 17 weeks and eight staff members to produce an outcome.

For a startup, the biggest risk was time. Building data infrastructure from scratch could burn a year or more before researchers saw real value. Manifold didn’t have that kind of runway, and neither did their customers. The team had watched that cost play out firsthand: one cancer center lost two years on a failed internal build before turning to Manifold. "Every delay in developing a new medicine is measured in lives. A month lost in analysis is a month patients wait for therapies that could help them," said Mohta.

Founded in 2016 as an AI consulting lab, Manifold pivoted to software in 2022 after seeing the same contradiction across the life sciences: scientists using cutting-edge tools were still managing data with spreadsheets, and a single research question could take months to answer. Manifold saw that agentic AI could fundamentally change how research gets done, on the idea that agentic AI would expand the life sciences market by orders of magnitude. Rather than replace scientific judgment, agents could remove the bottlenecks standing between a question and an answer, letting researchers direct strategy and validate results.

The opportunity required a foundation the team could trust and move quickly on. Manifold built directly on AWS data services, then extended upward, putting Anthropic’s Claude on Amazon Bedrock as the reasoning layer and turning its trusted research environment into an agent operating system. Just as important, working with AWS helped the team decide which services to build on and where Manifold’s own technology should live.

The goal was simple: get working agents into researchers’ hands fast enough to prove the concept and demonstrate value. If Manifold could show the UVA Comprehensive Cancer Center real results on one workflow, the door would open to an entire cancer center. And if agents could collapse timelines that had defined biomedical research for decades, the impact on patients would be hard to ignore.

To move from an idea on a napkin to a working product, Manifold made a deliberate choice: build the pieces that make the company unique, and partner with AWS for the foundation underneath. “We had a lot of focus on speed, getting to market quickly, so it often made sense to partner, working closely with the AWS team,” says Mohta. “We were able to understand how we should think about what services we want to build on, and then where the Manifold-specific pieces could fit in.”

That collaboration gave the team a clear architecture map. Working with AWS, Manifold built on data services such as Amazon S3, AWS HealthOmics, and AWS Glue to handle the foundational data work so engineers could focus on agentic capabilities and the life sciences–specific logic that sets Manifold apart. Because Manifold builds on AWS services rather than around them, onboarding that once took 12 months or more now takes three to four weeks.

With the foundation in place, Manifold assembled its agentic operating system: a trusted research environment that lets researchers pose questions, then agents execute the underlying multi-step work. Today, more than 300 researchers self-serve across a 130,000-plus patient DataMart, turning workflows that once required coordinated human effort into automated processes they can run themselves.

“Getting them to believe that our vision could be relevant across the entire cancer center, and then delivering on that, was super exciting,” Mohta recalls. Building on AWS meant the team never had to pick between moving fast and building something durable. The infrastructure was ready to grow as Manifold added workflows, brought on new research partners, and extended the platform.

The broader payoff is agentic capability that customers can build on. Manifold has raised $40 million across a Series A and Series B, with AWS named as a strategic collaborator in the Series B announcement, and customers are building on top of the agentic capabilities the team shipped. As Mohta puts it, the rewarding part has been “to see what we’ve been able to get to customers, and then the magical things they’re able to do on top of those capabilities.”

“With agents, we really have a unique ability to go a lot faster, and I think all of us need to be taking that opportunity, rethinking how we could apply this technology to ultimately get therapies to patients faster,” says Mohta. For a company whose mission is measured in the time patients have left, every workflow compressed from weeks to minutes is time given back.

Manifold AI went from a napkin sketch to an agentic solution that reaches millions of patients. If you're a founder with an idea that could change how your industry works, AWS AI can help you get from concept to working product faster. Explore the AI for Startups hub to access foundation models through Amazon Bedrock, accelerate development with Kiro, and apply for up to $200K in AWS Activate Credits to start building today.

AWS services used

AWS HealthOmics

AWS HealthOmics helps healthcare and life sciences organizations store, query, and analyze genomic and other omics data at scale, giving Manifold managed data services purpose-built for research workloads.

Amazon S3

Amazon Simple Storage Service (Amazon S3) is object storage offering industry-leading scalability, availability, security, and performance for the research data that fuels Manifold’s agentic workflows.

Amazon Bedrock

Amazon Bedrock is a fully managed service offering a choice of high-performing foundation models through a single API, along with the capabilities to build generative AI applications with security, privacy, and responsible AI—supporting agentic reasoning across Manifold’s platform.

AWS Glue

AWS Glue is a serverless data integration service that discovers, prepares, and combines data for analytics, helping Manifold move and transform research data across its agentic workflows.