Insilico Medicine announced on September 10, 2026, that the first patient had been dosed in GENESIS-IPF-3, a 52-week Phase 3 trial of rentosertib for idiopathic pulmonary fibrosis, a progressive scarring disease of the lungs. The drug brings two uses of AI into a late-stage clinical test: identifying a biological target and designing a molecule to act on it.
A separate finding has drawn the drug into a much bigger argument. An exploratory analysis of blood samples from its earlier trial found shifts toward younger-looking protein profiles. On Moonshots with Peter Diamandis, one panelist read those shifts as a glimpse of “longevity escape velocity”—medical progress adding life faster than time takes it away.
The clinical milestone and the longevity argument concern the same drug. They ask different questions.
A lung-disease treatment designed with AI
Idiopathic pulmonary fibrosis, or IPF, progressively stiffens the lungs and makes breathing harder; “idiopathic” means its cause is unknown. Existing treatments can slow decline, but do not cure the disease. The need for better options is urgent.
Rentosertib inhibits a protein called TNIK. Insilico used AI to identify that target, then generative AI to design a molecule intended to block its activity. The approach goes beyond searching a collection of existing compounds: it proposes new molecular structures for researchers to make and test.
Diamandis described that distinction enthusiastically on the podcast, while disclosing that Insilico is one of his portfolio companies. He also described Alex Zhavoronkov as a close friend and said he had been an early investor and adviser.
According to Insilico’s announcement, GENESIS-IPF-3 is a multicenter, randomized, double-blind study comparing once-daily rentosertib with placebo over 52 weeks. Participants are assigned by chance, and neither they nor the investigators assessing treatment know which assignment they received. The study is intended to test efficacy and safety over a longer period and in a larger group than the preceding 12-week trial.
Insilico describes it as the first Phase 3 trial of a generative-AI-driven innovative drug. That is the company’s milestone claim, not an approval. The lead investigator quoted in the announcement estimates that moving from Phase 3 initiation to regulatory approval could take three to four years under favorable conditions.
What the six aging clocks found
The study published September 7 in Nature Biotechnology examined a different question: could a disease-treatment trial also gather useful evidence about aging?
Researchers applied six proteomic aging clocks—statistical models that use patterns of blood proteins to estimate age—to samples from the earlier Phase 2a trial. A lower estimate means the measured profile looks younger to the model. It does not directly measure years of life remaining.
The original randomized, blinded trial enrolled 71 people with IPF in China during 2023–2024. Sixteen participants discontinued that trial. For the ancillary analysis, 43 consented and one lacked a required measurement, leaving 42 participants. Blood serum was sampled before treatment and at weeks 2, 4 and 12 across placebo and three dosing regimens.
The clocks indicated younger predicted profiles with treatment. Their greatest agreement appeared at week four, especially among participants receiving 30 mg twice daily. The podcast described shifts equivalent to roughly three to six years, with one panelist emphasizing a three-to-four-year change at the week-four peak.
The pattern did not keep advancing at that rate. Clock signals subsequently weakened or plateaued, despite sustained changes in the underlying proteins. A comparison with protein patterns in 55,319 UK Biobank participants found reversal of age-associated patterns with twice-daily treatment, but not with 60 mg once daily. Changes in lung function explained little of the variation in clock readings.
Fibrosis itself affects proteins used by the clocks. The researchers therefore could not separate a general effect on aging from changes associated with the lung disease. They called for testing in broader populations and validation against functional outcomes—whether people’s bodies actually work better, rather than simply producing different model estimates.
The dosing distinction also matters: the strongest clock consensus appeared with 30 mg twice daily, while the announced Phase 3 trial uses once-daily treatment. The earlier clock result cannot simply be carried over to the new trial.
From four weeks to “escape velocity”
The panel’s excitement came from comparing the treatment time with the apparent size of the clock shift. One panelist put it this way: “Four weeks of input, negative three to four years of output.” He argued that longevity escape velocity might already exist in isolated subpopulations, even if it had not arrived for everyone.
The reasoning was that AI could produce a succession of narrow advances: a treatment helps one group, another helps a different group, and their cumulative effect eventually outruns aging. He acknowledged the limited population and clinical setting, but treated the clock result as an early example.
That argument depends on translating a younger protein signature into additional life. The exploratory analysis did not establish that translation. It did not show that participants had gained three or four years of survival, that their bodies had undergone equivalent rejuvenation, or that repeated treatment would keep producing gains. Agreement among six clocks is useful supporting material for further investigation, but all six were examining protein measurements from the same small trial—not six independent demonstrations of longer life.
What would count as an aging treatment?
The discussion turned to regulation. A panelist described recognition of aging as a condition and treatment target by the US Food and Drug Administration as the biggest obstacle. Participants floated specialized regulatory regions that might offer faster development and access; another called the idea “medical charter cities.” These were proposals, not announced approval pathways for rentosertib.
An aging-treatment pathway would still need an endpoint: a defined outcome showing that a treatment helps patients. This study explores whether protein clocks could contribute to that assessment alongside disease-specific testing. It does not establish that a clock shift can substitute for evidence of better health or longer survival.
Rentosertib’s current route is more concrete: demonstrate safety and benefit for people with IPF. Near the end of the podcast segment, one panelist explained the personal stakes: “My dad’s actually got IPF, so I can’t wait for this drug to hit the market.” The next step toward that hope is the longer controlled trial—not a countdown to approval, but a test of whether the treatment can deliver sustained benefit.