Kebotix raises $11.5 million to automate lab experiments with AI and robotics
Kebotix, a startup developing AI tools to expedite the discovery of chemicals, today announced that it has raised $11.5 million. Investors were evidently impressed by the company’s machine learning and robotic process automation suite, which enables Kebotix’s private and public sector partners to uncover materials faster than with manual techniques.
Chemical development is a complex process that requires weeks (or months) of labor and plenty in the way of capital. For instance, the National Institutes of Health’s National Center for Advancing Translational Sciences (NCATS) has an average experimental time of 49 hours, and companies like Koura spend millions and even billions of dollars refining environmentally friendly materials. Kebotix’s products promise to cut down on the workload and expense, in part by combining data with evolutionary AI algorithms and autonomous machines.
Kebotix, which works out of MIT-affiliated VC firm the Engine and whose technology was developed at the Harvard lab of Alán Aspuru-Guzik (now a researcher at the University of Toronto), offers robotic arms and AI models that learn statistical representations of compound properties. Its self-driving systems can dip pipettes into dishes and transfer liquids into other machines that test their optical properties, while its models analyze the results and formulate hypotheses.
The models in question can iterate on well-understood, desirable compounds to come up with closely related new examples. Alternatively, they’re able to isolate and ditch molecular designs that aren’t likely to be useful.
The results of each experiment Kebotix conducts are fed back into the system so that it continuously self-improves. In this way, a library of thousands of material candidates that might lead to new products can be evaluated. From there, a set of validated probabilistic models can be generated to predict the properties of promising molecules.
In a pilot program involving the NCATS, Kebotix says its algorithms found optimal conditions and assay performance with only 55 measurements (up to 20 of which ran simultaneously), allowing it to conduct 294 experiments with greater than 95% accuracy. This amounted to a fivefold reduction in costs for lab supplies and runtime, from 49 hours to around 9 hours.
Kebotix intends to focus on molecule discovery for electronic applications and then new polymers and alloys, but it believes its system will eventually uncover compounds that absorb pollution, combat drug-resistant fungal infections, and serve as more efficient optoelectronic components. In a step toward this vision, the startup recently announced that it will work with chemists at the Northeastern University’s Lopez Lab to accelerate the development of multicolor chromophores used in cancer surgery and light-activated therapy.
Kebotix isn’t the only party investigating AI-assisted lab automation. MIT researchers last year detailed a robot that autonomously performs fluid dynamics experiments, observes the results, and plans a follow-up. Elsewhere, a team at the University of British Columbia developed Ada, a robot that can mix different solutions, cast them in films, perform other processing and testing steps, and log the results. And the coauthors of a new Rutgers study claim to have created a blood-sampling robot that performs as well or better than people.
Perhaps unsurprisingly, in a recent survey of 100 pharmaceutical executives, Pharma IQ found that 94% believe intelligent automation technologies like robots and AI will have an impact on lab practices within two years.
Life science investor Novo Holdings led Kebotix’s series A, which comes after a $5 million seed round in August 2018 that saw participation from One Way Ventures, Flybridge Ventures, Baidu Ventures, Embark Ventures, and Propagator Ventures. Kebotix says it will keep the round open another 30 days for additional participation “due to significant interest.”
Kebotix has 17 full-time employees and says it is currently generating revenues with its partnerships (including one involving energy giant BP). With the new funds, the company plans to scale operations and hiring and accelerate R&D and product development.