What challenges did PozeSCAF Discovery Solutions face before AI implementation?

Older GROMACS versions caused scalability and performance issues. Molecular dynamics simulations took more than 30 hours, slowing hit identification and hit-to-lead cycles. Higher compute costs and slower preclinical timelines hurt competitiveness.

What was the impact of the AI solution?

Cut simulation runtime from 30 hours to under 15, a reduction of more than 50%. Increased throughput to about 2.5 times more simulations in the same timeframe. Reduced compute costs by 25–30%. Saved an estimated 2–3 months during the preclinical phase.

Source

ExploringEvidence: Medium65/100

PozeSCAF Discovery Solutions — Amazon Bedrock + high-performance simulation on AWS to accelerate drug discovery

Use case typeLife sciences innovationUpdated Jun 13, 2026

PozeSCAF Discovery Solutions (formerly Immunocure Discovery Solutions) turned to AWS for scalable, high-performance infrastructure to optimize molecular dynamics workloads. The company cut simulation runtimes by more than half, reduced compute costs, and accelerated its drug discovery pipeline. It also began exploring generative AI/agentic workflows with Amazon Bedrock to build knowledge graphs from project data and flag potential side effects earlier.

Organization: PozeSCAF Discovery Solutions
Industry: Pharma
Location: United States
Published: May 2026

Reported outcomes

Simulation throughput increase: Approximately 2.5×

Productivity & throughput

Preclinical time saved: 2–3 months

Catalog median for productivity & throughput deployments: +40% across 100 reported metrics. Compare benchmarks →

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Simulation runtime reduction: 50% decrease

AWS Solutions Case StudyMay 27, 2026Case studyExplicit claimMedium evidence strength

reduced simulation runtime by more than 50 percent—from 30 hours to under 15

Normalized claim

Simulation throughput increase: 2.5 x increase

AWS Solutions Case StudyMay 27, 2026Case studyExplicit claimMedium evidence strength

teams now run about 2.5 times more simulations in the same timeframe

Normalized claim

Compute cost reduction: 25-30% decrease

AWS Solutions Case StudyMay 27, 2026Case studyExplicit claimMedium evidence strength

reduced compute costs by 25–30 percent

Normalized claim

Preclinical time saved: 2-3 months decrease

AWS Solutions Case StudyMay 27, 2026Case studyExplicit claimMedium evidence strength

this saves approximately 2–3 months

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: PozeSCAF Discovery Solutions
Provider: AWS
Maturity: Exploring
Linked source: AWS Solutions Case Study

It also began exploring generative AI/agentic workflows with Amazon Bedrock to build knowledge graphs from project data and flag potential side effects earlier

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1HPC optimization
2Drug discovery acceleration
3Generative AI exploration

Benchmarked Amazon EC2 instance types including GPU and HPC options.
Fine-tuned GROMACS parameters using GPU acceleration and upgraded to the latest version.
Used a Slurm cluster on AWS for large-scale compound screening and explored Amazon Bedrock for knowledge graphs and agentic AI workflows.

Technologies

Amazon EC2 AWS ParallelCluster AWS Batch Amazon Bedrock Slurm

Cut simulation runtime from 30 hours to under 15, a reduction of more than 50%.
Increased throughput to about 2.5 times more simulations in the same timeframe.
Reduced compute costs by 25–30%.
Saved an estimated 2–3 months during the preclinical phase.

Architecture

PozeSCAF optimized molecular dynamics workloads on AWS by benchmarking Amazon EC2 GPU and HPC instances, tuning GROMACS parameters for GPU acceleration, upgrading the software stack, and running large-scale screening on a Slurm cluster on AWS. The article also notes early exploration of Amazon Bedrock for knowledge-graph and agentic workflows.

Sources & evidence1

Evidence: Medium65/100Evidence strength