James Field believes his automated laboratory is closing in on a powerful new drug to kill cancer cells with unprecedented precision. Yet the drug wasn’t created by any of the biologists working in his lab — it was generated by artificial intelligence.

Field is a protein engineer working in synthetic biology, a discipline that uses the latest technology to engineer novel cells or biological components. In his world, researchers often talk about a cycle known as DBTL — design, build, test, learn — an iterative process where the end product, whether it is a molecule or a new strain of bacteria, is repeatedly optimized until it has the traits scientists seek. Field’s company, LabGenius, is one of a group of startups that is at the forefront of a revolution in which AI increasingly runs the DBTL cycle from beginning to end.

The work takes place at the company’s sophisticated automated laboratory in a former biscuit factory in south London. Such laboratories, known as biofoundries, first emerged in the 2010s as gene editing technology, combined with “high-throughput” machines — which can conduct hundreds of experiments at once — accelerated the DBTL cycle by many orders of magnitude.

Experiments using biofoundries often involve adapting a piece of DNA that is then added to a host organism — frequently the bacterium Escherichia coli or yeast — to get it to produce a desired molecule, which can be anything from an antibody to an ingredient for sustainable plastic.

Algorithms have been integral to biofoundries from the start in managing the operation of high-throughput machines. But the industry is now approaching a point where AI models are becoming sophisticated enough to program DNA to produce new medicines and industrial chemicals, an emerging field known as SynBioxAI.

Rather than helping only to run the machinery, AI can now work iteratively, learning from results and suggesting new avenues of experimentation, reducing the development time for new drugs or industrial chemicals from years to months.

According to Paul Freemont, who heads synthetic biology at Imperial College London, we are approaching a powerful point of “convergence of automation, data and machine learning and AI.” He was the founding chair of the Global Biofoundries Alliance, an international group of publicly funded biofoundries that launched in 2019 in Kobe, Japan, with 16 members. The group has now grown to over 40 members across the world, spread from China to Mexico.

Since that launch, countries have been rapidly investing in biofoundry capacity. The US, for example, is spending $75 million on the development of five new biofoundries, while China has named biomanufacturing and synthetic biology as core strategic priorities for its 2026-2030 five-year plan.

The objective is to engineer biology, programming cells to produce new drugs and industrial products, a technology that Freemont says is going to “become essential for mankind.”

However, he believes the field hasn’t fully reached this point of convergence. “I think there’s a massive amount of technical work that needs to be done to achieve the vision of what people are trying to do,” he says.

One key technical task is the work of standardization. In 2025, Freemont and colleagues in the United Kingdom, South Korea and the United States published a paper proposing a way for biofoundries to achieve this. “The idea was to give us a lexicon, if you like, a sort of standardized nomenclature, standardized workflow,” he explains.

The Korea National Biofoundry and the London Biofoundry — housed in Freemont’s department at Imperial College — are among nine organizations taking part in a project to develop such global standards and metrics. Based at the University of Illinois Urbana-Champaign, the effort was launched in 2024 with funding from the US National Science Foundation, with the goal of making genetic design “predictable, reproducible and scalable.”

Another hurdle to be overcome is the sheer complexity of biological data. In a 2025 report on the future of synthetic biology, AI and automation, the Organization for Economic Cooperation and Development said that one of the key trends in the sector was the need to aggregate these data — “addressing fragmentation, promoting high-quality curated datasets with long-term sustainability, and balancing openness with security.”

One example of the messiness of biological data right now is in plant research, where genomes are “scattered around different databases” and “of varying quality in different formats,” says plant biologist Anne Osbourn of the John Innes Centre, a plant science, genetics and microbiology research institute in Norwich in the United Kingdom. “They’re not standardized. So that presents a challenge.” To proceed with their work, her team had to develop a way to assimilate all of the publicly available plant genome sequence information so algorithms can search it to identify the genes they were interested in.

In 2024, her group reported reconstituting the full, 20-step process involved in making a drug called QS-21, sourced from bark of a tree in Chile, in a tobacco plant. QS-21 is used to boost the effectiveness of vaccines; now, through a startup called HotHouse Therapeutics, Osbourn and her colleagues are working to create more and better drugs of that kind.

Plant genomes are a trove of genes that could be mixed and matched to create next-generation drugs, Osbourn says — the potential, she adds, is “massively disruptive.” But right now, very few genomes have been sequenced, and even fewer analyzed. “We don’t even know the vast majority of the chemistry that’s out there,” she says.

Another biofoundry, at the Earlham Institute, situated in the same research park as the John Innes Centre, is developing a way to create a plant artificial chromosome for potatoes — scientists could efficiently add genes to it to make the plant produce additional nutrients or become more resistant to pests. The current objective is to build a library of potato control elements that could be pieced together on the synthetic chromosome to introduce desired traits.

AI is helping the researchers know where to look. Without it, it would be a gargantuan task — a potato has tens of thousands of genes to choose from, and each gene has a control element at the beginning of the gene and at the end, and these can be combined in different ways.

“Your experimental space is very large,” says Melissa Salmon, operations manager at the Earlham Biofoundry. “The role of AI would be to help us efficiently test our space.”

Machine learning helps researchers distill the massive amounts of data gathered from sequencing genomes into a small subset, which the scientists then use to design their next experiment. And the automated equipment of the biofoundry allows them to test multiple variations at the same time. “I think it has changed the science field massively,” says Salmon.

Full end-to-end AI automation of the DBTL cycle is often the ultimate goal of biofoundry development, yet biological complexity can get in the way. This is especially true when the output is a whole organism, explains Michael Köpke, chief innovation officer at LanzaTech, an Illinois-based company. It specializes in using bacteria to convert CO₂ and other waste gases, including emissions from steel plants and gasified biomass or municipal waste, to fuels and raw materials for clothing companies such as Adidas and H&M.

His company is using a biofoundry to make and screen thousands of new bacterial strains every week. They’re optimizing the microbes to be more efficient at converting CO₂ and producing ethanol or making other target molecules. These can be used for fuel and manufacturing, in the quest for an economy that is better at recycling carbon.

At the moment, Köpke’s team is using AI in three parts of the DBTL cycle. AI is helping to design new enzymes, to build the required parts of DNA, and to interpret the results of tests (the “learn” part of DBTL). “We have not a fully autonomous loop yet,” he says — right now, scientists must still take these suggestions and start the next test cycle. But he believes the whole process might become automated in the future with some minimal human oversight.

AI is taking a critical role in the design part of the cycle in Field’s fully automated lab, too, unlocking new designs that would not have occurred to a human because of the sheer number of parameters AI can take into account at the same time — things like the functional properties of the antibody building blocks, the geometry of the building blocks in the final molecule, and biophysical properties like clusters of positive and negative charges.

AI can then integrate all of these data to predict the molecule’s function and whether it can be feasibly developed. As a result, says Field, “what it's able to do is to wander into areas of design space that humans would never normally have looked in.” Perhaps not surprisingly, he adds, “the molecules that come out of our discovery process — they look really weird.”

One promising result of this is LabGenius’ lead product, an antibody called LGTX-101, a drug that can trigger the immune system to kill cancer cells selectively, sparing healthy cells in the process. Unlike similar drugs on the market, LGTX-101 works by using three binding components that recognize Nectin-4, a protein that exists on many healthy cells but is present in greater numbers on certain tumor cells.

Because LGTX-101 needs to attach to Nectin-4 at three points before bringing in an immune cell to finish the job, it is over 70,000 times more likely to target a cancer cell than a healthy one in tests, the company says. It has already been shown to be effective at killing tumors in mice transplanted with human breast cancer cell lines; the company will begin early-stage human trials in 2027.

Even researchers who don’t have the budget to build their own biofoundry can get in on the action: US-based firm Ginkgo Bioworks operates a lab-for-hire model from its Boston-based Nebula biofoundry.

The biofoundry is fully automated, using AI and over 100 robots to design, execute and analyze experiments. “It has 70 different scientific devices on it. And the whole thing is controlled by software,” says Ginkgo chief executive Jason Kelly. “And so, if you submit a software order into that laboratory, your experiment will get executed.”

The service has now been extended to online users through Ginkgo’s Cloud Lab service, launched in March this year. “The idea,” explains Kelly, “was, could we make that also accessible to people outside the building to order an experiment?”

The Cloud Lab works through an online AI agent called EstiMate, allowing scientists to submit their experiment in human language and receive an immediate assessment and price. It’s a way of opening up the technology to scientists around the world who might otherwise not be able to access an automated biofoundry, Kelly says.

In fact, Kelly sees a future in which an AI scientist could allow even nonscientists to run experiments. “They could interact with an AI scientist who would design the experiments for them that would run on the autonomous lab — and now they can do science,” says Kelly.

But flinging open the doors to this technology also poses a risk, says biosecurity specialist Leyma De Haro, a scientist at Colorado-based engineering company Merrick and author of the book Biosecurity in the Age of Synthetic Biology. It means that people will have access to the technology in countries that lack regulations to prevent them from using it for nefarious purposes, such as making a bioweapon. “That makes the entire world vulnerable, not just that country or that region,” she says.

De Haro also warns that biofoundries might be vulnerable to hacking, especially when they’re managed by AI. An AI agent controlled by a hostile government or terrorist group, “could... hijack your system without your knowledge just to sabotage your experiments, which in itself could be damaging,” she says. “They could steal your data or your information for their own purposes, or they could try to hijack your system to do experiments for them[selves] without your knowledge.”

The problem is that the technology is moving at such breakneck speed that regulators are struggling to keep pace. “You have to think faster also about the ethical issues of what this could mean,” says Marileen Dogterom, a biophysicist from the Delft University of Technology in Netherlands, who was the co-coordinator of a report by UNESCO’s International Bioethics Committee on the ethical issues surrounding synthetic biology.

Noting that synthetic biology is increasingly propelled by AI, the report made a number of recommendations so that the field can develop responsibly, such as setting up a global steering body and a monitoring organization and establishing ethical guidelines. In the hope of spurring action, the group is presenting the report at meetings where scientists, governments and companies come together, such as the Synthetic Biology & Bioethics Symposium jointly organized by the UNESCO Turkey National Commission and Acıbadem University in Turkey in February 2026.

“The urgency,” says Dogterom, is “to think about the risks, the biosafety, the ethical issues, what to do with this new technology — how to make sure it benefits the planet as a whole and not a very small part of it.”

Editor’s Note: This article was updated on August 19, 2026, to correct the description of LabGenius’s location. Its automated lab is sited in a former biscuit factory, not a former brewery.

This article originally appeared in Knowable Magazine on August 18, 2026. Knowable Magazine is an independent journalistic endeavor from Annual Reviews, a nonprofit publisher dedicated to synthesizing and integrating knowledge for the progress of science and the benefit of society. Sign up for Knowable Magazine’s newsletter.