Fred Parietti and Multiply Labs: Transforming Drug Manufacturing Through Robotics
Fred Parietti built an early robotics prototype in his Cambridge apartment. A decade later, Multiply Labs is tackling a much bigger problem in medicine. How do you make more doses when each patient’s treatment may require its own production run?
A robot can lift a metal component off an assembly line thousands of times. Ask it to connect two pieces of flexible tubing without compromising sterility, and the job looks different. The tubing bends. The connector has to line up. The motion has to work each time. In 2026, Multiply Labs and CPC Biotech described their work on a sterile connector with a rigid interface that a robotic gripper could handle. A small component had become part of solving a much larger manufacturing problem.
That detail helps explain what Fred Parietti, co-founder and CEO of Multiply Labs, is building. The San Francisco company develops robotic systems that operate the instruments used to make advanced therapies. Its approach starts with the work people already do inside a manufacturing facility and asks which steps a machine can perform consistently. Answering that question can involve redesigning a connector, testing how cells respond, or figuring out how the new equipment will fit on a production floor.
The Idea Began With a 3D-Printed Capsule
Parietti was working on robotic limbs at MIT when Alice Melocchi, a visiting doctoral researcher in chemical engineering, saw another use for the 3D printers in his workspace. She had been exploring ways to make personalized drug capsules and saw how printing could help create capsules with different compartments. Parietti, who earned his doctorate in mechanical engineering from MIT in 2016, built an early robotic prototype (in the kitchen of his Cambridge apartment).
MIT’s 2016 coverage identified four co-founders of Multiply Labs: Parietti, Melocchi, Tiffany Kuo, and Joe Wilson. The company joined Y Combinator that summer and demonstrated custom supplement capsules that released different ingredients at different times. The capsules helped the team test whether a single machine could produce small batches to different specifications.
The team later saw an application in pharmaceutical development. A company running a clinical trial may need several strengths of the same drug as it investigates which dose to use. Multiply Labs developed robotic systems to make small batches of capsules for pharmaceutical companies. By 2021, it was working on a different challenge with Cytiva and researchers at the University of California, San Francisco: automating parts of cell therapy production. That meant learning to handle living cells and a new set of instruments.
Why a Patient-Specific Therapy Is Hard to Make at Scale
Some CAR T-cell therapies begin with immune cells collected from the patient who will receive the treatment. At the manufacturing facility, specialists modify the T cells to recognize a specific target on cancer cells. The lab grows more of the modified T cells and sends them back to the hospital. The patient then receives the cells by infusion. In this form of therapy, known as autologous therapy, the starting material and the finished treatment belong to the same patient.
Consider what happens when a facility needs to serve twice as many patients. Making one larger batch will not solve the problem when each patient needs a separate production run. The facility needs room and equipment to handle more runs, staff who can oversee the work, and a reliable way to keep each patient’s material correctly identified. The challenge is to run more batches side by side while maintaining control over each one.
This is where Multiply Labs sees a role for robotics. Multiple runs can be at different stages at the same time. One group of cells may be growing in a bioreactor while another is ready for sampling. A robotic system can coordinate timed tasks across those runs. How much extra capacity a facility gains will depend in part on which step is holding up production. Faster transfers help most when transfers are the constraint. If the limit is the number of instruments available to grow cells, the facility needs to address that too.
The Robots Work With Equipment Manufacturers Already Use
Multiply Labs builds enclosed systems it calls robotic clusters. Each cluster includes robotic arms, software, and biomanufacturing equipment. The arms handle vials, syringes, and bags. The software tells the robots what to do next and logs data as they work. The company designs the clusters to work with compatible instruments manufacturers already use, including equipment for growing, counting, and processing cells.
The design choice matters because an instrument can work well in a manual process and still be difficult for a machine to operate. A person can feel a connector shift or adjust their grip on a soft bag. A robot needs a repeatable way to pick up the object, position it, and complete the movement.
For an operations leader, working with familiar instruments can make adoption easier to assess. The team can start with a process it understands and examine what changes when a robot performs a step. For a quality leader, the same project raises a different question. What evidence will show that the new method preserves the product characteristics that matter?
What the Research Shows So Far
In a 2024 study in Cytotherapy, researchers tested a prototype cluster against a manual process for growing a type of T cell. They reported comparable results for the number of cells produced, the proportion that stayed alive, and the measured characteristics used to identify them. Multiply Labs funded the work, and industry partners contributed equipment and funding. The comparison gives manufacturers evidence about how the prototype performed on those measures under the study conditions.
A later study compared the cluster with skilled operators on individual tasks, including transferring cells between sterile bags, counting cells, and drawing liquid from a vial into a syringe. The researchers reported equivalent performance on the measures they selected for those tasks. Their work shows how a company can test a complex system piece by piece, with a defined comparison for each operation.
The economics come from a separate comparison. In a case study published by robot supplier Universal Robots, research collaborator Jonathan Esensten reported an approximately 74% reduction in manufacturing cost when a particular manual process was compared with its robotic counterpart. In the same case study, Parietti estimated that the system could produce up to 100 times as many patient doses per square foot of cleanroom space. One figure describes a cost comparison for a particular process. The other estimates how much the system can produce per square foot of floor space. A manufacturer evaluating the system would need to see how its own workflow, facility costs, and equipment use affect both measures.
From a Working Prototype to Commercial Production
In October 2026, Multiply Labs announced a $75 million Series B, bringing its total funding to more than $100 million, according to the company. Patrick Soon-Shiong led the round through NantWorks, with participation from investors including AstraZeneca and Lingotto. Multiply Labs said it planned to increase manufacturing capacity as it moved from clinical-stage deployments to commercial-scale production. The company also planned to expand its engineering, regulatory, and commercial teams.
That next stage puts the technology in a different setting. Before using the system in commercial production, manufacturers need answers to three questions. What happens to a batch if a machine stops mid-run? Who reviews the records if a process step falls outside its acceptable range? How soon can production restart? Multiply Labs says its customers own and operate the systems, keeping production and manufacturing data in-house. That puts decisions about how the robots fit into daily operations (with the teams responsible for making and releasing the product).
The evidence needed to assess a manufacturing change depends on the therapy and the process. FDA guidance for CAR T-cell products discusses comparing material made before and after a change. Its November 2026 guidance on potency for cell and gene therapies also addresses how manufacturers can build controls and testing into production to help verify that a product has the intended biological activity. For executives and board members, that makes the quality team’s assessment part of the business case from the start.
Parietti’s path from a kitchen prototype to an enclosed manufacturing system illustrates how much of industrial innovation happens in the details. In one project, the work came down to giving a robot a reliable grip on a connector.