← Back to Home

Bringing AI Designs into Experiments Faster: Telesis Bio Licenses Enzymatic DNA Synthesis Platform

Digital antibody designs still have to withstand real-world experiments. Telesis Bio has signed licensing agreements with multiple AI drug discovery companies in an effort to shorten the wait for DNA; however, daily output and a 36-hour experimental cycle remain capabilities projected by the company.

By SURL BioNews

AI can rapidly propose antibody designs, but laboratories must first turn those sequences into DNA before they can produce antibodies and test their performance. This wait between computer and lab bench is the step Telesis Bio hopes to shorten. On August 25, 2026, the company announced that it had signed commercial licensing agreements with multiple AI-focused drug discovery companies to bring the Gibson SOLA enzymatic DNA synthesis platform into customers’ laboratories. Customer identities and financial terms were not disclosed.

The platform targets the iterative process of antibody discovery: researchers produce DNA constructs for antibody expression based on their designs, obtain experimental results, and then feed the data back into models to refine the next round of designs. Related coverage by Lab Manager also discusses in-house DNA synthesis within this “design, build, test, learn” cycle. If DNA supply cannot keep pace, even a model generating numerous ideas must wait for materials to arrive before those ideas can be tested.

According to Telesis Bio’s announcement, Gibson SOLA combines AI software, fully enzymatic ready-to-use reagents, and a modular DNA synthesis workflow that can operate with standard automated liquid-handling equipment. The agreements also include solution design and implementation support, meaning the licensing arrangements involve integrating synthesis steps into customers’ existing laboratory workflows, as well as supplying reagents.

CEO Eric Esser said customers adopting the platform are expected to produce hundreds to thousands of antibody expression constructs per day, with the time from an AI-generated hypothesis to experimental results potentially reduced to as little as 36 hours. A same-day report by S&P Capital IQ published on MarketScreener also recorded these figures. However, the announcement describes what customers will achieve after implementation and does not include measured data from named laboratories. The 36-hour figure should therefore be understood as the projected fastest cycle, rather than an established norm demonstrated across all types of antibody experiments.

Publicly available information also does not specify the models, training datasets, or antibody targets these customers use, or provide comparative results for synthesis success rates and costs. This leaves outside observers unable to assess whether shorter production times can consistently yield better antibody candidates. The number of DNA constructs produced each day measures capacity to produce experimental materials; it does not directly equate to the number of molecules with therapeutic potential obtained.

These licensing agreements provide a concrete commercial signal: AI drug discovery companies are beginning to incorporate DNA production into their own laboratory workflows, hoping to obtain biological data for model learning more quickly. The measures that can test this promise will be quality, cycle times, and candidate molecule performance as customers sustain operations. The announcement has yet to provide clinical validation or drug approval results. Multiple stages of development still separate faster experiments from actual improvements in treatment.

References

  1. Lab Manager
  2. Telesis Bio
  3. S&P Capital IQ via MarketScreener