Introducing Science Foundries, a new way for scientists and agents to work together
Today we’re launching Science Foundries, a new way to run scientific processes end-to-end with agents and scientists working as a team.
Build a Science Foundry in two steps. Pick a process or goal: produce proteins, optimize a promising lead, or improve a manufacturing process. Then set who does what: which steps agents run and when scientists need to approve. From there, the Foundry runs. Agents execute and monitor work, coordinate handoffs, and surface key decisions to scientists, all working from the same scientific data and context.
We’ve spent the past eighteen months bringing AI to science. Our agents are the fastest growing capability we’ve ever released. Scientists delegate real work to them daily. But that’s a starting point. Rebuilding entire scientific workflows around agents will be much bigger.
For decades the number of programs a company could run was set by headcount, lab space, and budget. That is changing. Big pharma, big tech, and AI-native biotechs are investing in autonomous labs and moving more work to CROs. Models produce more candidates and agents work around the clock. A three-person team can now have the productive capacity of thirty, but only if they coordinate work with their agents, and across internal and external labs.
Science Foundries power this new way of working.
Example: A Science Foundry for high-throughput biologics
Over the last few months, we embedded with scientists to build Science Foundries across R&D. One example was for a pharma team that’s designing, producing, and testing antibodies at high throughput.
The team has models that design sequences, CROs that synthesize constructs, and lab automation for production and assays. Managing this process typically requires scientists to manually process data and feed it from one step to the next. Their Science Foundry now automates this process end-to-end:
A design agent uses scientific models to generate and evaluate sequences.
A cloning agent drafts the assembly strategy and places orders to a CRO.
A lab agent dispatches instructions to workcells, captures the data, and performs QC.
An analysis agent interprets the results and updates training datasets.
A Science Foundry delivers an agentic workflow connecting models, CROs, lab automation, and scientists.
With Benchling Agents capturing structured data along the entire process, scientists can spend more time on the decisions that matter: what design strategies to focus on, which candidates to advance, how to optimize the assays.
Less whitespace between each step also speeds up science. Assays finish at 11 PM and agents automatically pick up the results, presenting an analysis by the time the scientist starts their day. Multiply each experiment starting a day sooner and an R&D program can speed up by months.
Parabilis Medicines: From individual agents to Science Foundries
Parabilis Medicines, a clinical-stage biopharma pioneering a new class of alpha-helical peptides, has experienced the progression from chat, to standalone agents, to Science Foundries.
“AI and automation are transforming every part of the drug discovery process. We are working to empower lean teams of scientists by creating tools that handle as much of the execution as possible, and Science Foundries in Benchling are the perfect complement to automated labs - with teams of agents that keep science moving around the clock by automating the data capture, analysis, and handoffs needed to turn one experiment into the next.”
— Parabilis Medicines, John McGee, Ph.D. Scientific Co-Founder and Head of Platform
Parabilis’ lean scientific team is running several programs in parallel. They started out using individual Agents for discrete tasks like defining candidate-target profiles, experimental design, and SPR analysis.
Now, they’re deploying Science Foundries that bring teams of agents together across entire scientific processes like protein production, screening, and lead optimization—compressing whitespace between experiments and helping scientists spend more of their time on the decisions that drive programs forward.
Build your first Science Foundry
We’re building Science Foundries for teams across R&D:
Therapeutic Areas: Search literature to help define target candidate profiles, suggest assays to reuse or tweak from past programs, and design and orchestrate assay cascades.
Preclinical: Review in vivo study designs, process data from CROs, present analyzed results, and write study reports
Process Development: Propose DoE, process data from bioreactors, orchestrate analytical requests, and write study documentation
If you’re already using Benchling, you can set up a Science Foundry today. Pick a process that is key to accelerating the entire organization: protein production, screening, assay development, method development, or in vivo work. Then start a conversation with Benchling AI and have it build your first Science Foundry.
Reach out to us here for a free consultation on deploying Science Foundries for your organization.
