Benchling accelerates scientific discovery with Claude in Amazon Bedrock

Benchling, a leading R&D platform for biotech, uses Claude in Amazon Bedrock to embed AI agents directly into scientific workflows, automating tedious tasks and accelerating innovation for researchers working with complex biological data.

With Claude, Benchling:

• Saves scientists up to 2 weeks spent transforming complex data with Data Entry Assistant
• Improves data quality and experiment reproducibility with Notebook Check
• Speeds up answering scientific questions from hours to seconds with SQL Assistant

Building AI into everyday scientific work

Building effective AI in life sciences is uniquely challenging. Biological data is fragmented, highly specialized, and riddled with complexity. Scientific workflows are nuanced and often opaque to machines. Intellectual property concerns loom large. Yet the potential is enormous—AI models that can analyze scientific data, draft regulatory filings, or generate novel hypotheses to suggest new research directions.

This is precisely the environment that Benchling was built for. As the R&D platform for biotech, Benchling is embedded in scientists' daily work—managing data, documenting experiments, running analysis, and powering collaboration. Now, with Claude in Amazon Bedrock, Benchling is embedding AI agents directly into these workflows to automate tedious tasks, speed up innovation, and set a new standard for AI in science.

"Up to 25% of scientists' time is spent on capturing and aggregating data," says Ashu Singhal, Benchling's co-founder. "Time that could be spent on actual science." Whether at universities or advanced biopharma companies, researchers still transcribe results by hand, aggregate data in spreadsheets, and meticulously check analyses line by line—all tasks that could be automated with the right AI solution.

Why Benchling chose Claude in Amazon Bedrock

Benchling selected Claude in Amazon Bedrock after rigorous evaluation using domain-specific benchmarks with real scientific data and workflows. Claude 3.7 Sonnet consistently outperformed other models in accuracy, flexibility, and comprehension of scientific content.

"When we tested various models on complex scientific documents, Claude 3.7 Sonnet could understand and extract data from complicated file formats. It handled complex transformations that other models simply couldn't," said Singhal.

Security was equally important for Benchling's customers. Life science companies trust Benchling with their most valuable intellectual property—molecules and their associated data. Since Benchling is built on Amazon Web Services (AWS) infrastructure, using Claude in Amazon Bedrock provided essential security advantages by keeping AI within the AWS secure environment.

Singhal explained, "The ability to use models that stay within the AWS cloud, that are secured the same way as the rest of our product, dramatically increases our customers' trust in adopting AI." This seamless security approach addresses the data protection concerns that have historically made life sciences companies hesitant to use new AI services for their sensitive data.

How Claude transforms scientific workflows

Benchling has launched three AI assistants powered by Claude, each targeting a critical pain point in scientific research:

• Data Entry Assistant: Transforms unstructured data from study reports or instrument files into structured, searchable records. This saves hours per study, reduces transcription errors, and converts disconnected documents into usable data. "Almost always in these external study data transcription processes there's some data error that happens," explained Singhal. By automating this process with Claude, Benchling significantly reduces these errors.
• Notebook Check: Automates the initial review process for scientific documentation, identifying errors and missing data before human reviewers see them. Over 100 organizations use this feature to accelerate reviews while catching mistakes that might otherwise go unnoticed.
• SQL Assistant: Enables scientists to generate database queries and dashboards using plain English. This helps teams, especially those with limited technical expertise, unlock insights buried in their data.

Elevating research quality and accelerating discovery

Benchling's Claude-powered assistants are transforming how scientific research happens. Beyond simply saving time, these tools fundamentally improve research quality by reducing errors and standardizing data. For example, by automating manual data transcription—a process notorious for introducing errors—Data Entry Assistant helps prevent mistakes that would otherwise cascade through the research process.

"So much of science is about answering questions off of data," said Singhal. When information is automatically organized in standardized formats, researchers can investigate relationships that were previously hidden in disconnected documents, such as the exact correlation between drug formulation and effectiveness across multiple studies.

The impact extends beyond individual scientists to entire organizations. With standardized data formats, better documentation, and automated quality checks, teams can collaborate more effectively and build on each other's work with greater confidence. This creates a compound effect: not only do individual experiments move faster, but the collective knowledge of the organization grows more rapidly and reliably.

By removing the friction of manual workflows and scattered data, Benchling and Claude are helping research teams focus on what they do best: advancing scientific discovery.

Building a future where AI accelerates scientific breakthroughs

Benchling envisions a future where AI fundamentally changes how science gets done. Their vision extends beyond automating routine tasks to creating AI systems that can design experiments, guide data collection, and optimize protocols in real time.

"Scientific use cases will push AI models in unique ways," says Singhal. "These are some of the hardest reasoning problems out there—and solving them will define the next generation of AI."

By working with Anthropic and AWS, Benchling aims to build AI systems that truly understand the scientific process and can contribute meaningfully to research. The goal isn't to replace scientists but to amplify their capabilities—helping them make connections across vast datasets, generate novel hypotheses, and ultimately accelerate the pace of discovery.

Benchling's mission reflects the urgency of scientific advancement. By transforming how researchers interact with their data and enhancing scientific decision-making, Benchling and Claude aim to help more life-changing treatments reach the patients who need them most.