Nicolas Tilmans

I recommend working with @nathanbenaich to anyone, really a great supporter of real AI solving real world problems. Congrats to @airstreet and Nathan!

@ladanuzhna It’s true that BD alone isn’t sufficient. It is necessary, as is a person who *really* wants to do the deal on the buy side. Usually that’s from the science org, and usually has to be very senior for it to go through. Bad BD will definitely kill deals despite a champion.

My working theory of pharma partnerships is that a (biotech) BD person talking to a (pharma) BD person never closes any deals. In all the big partnerships, it's clear that a scientist on one team talked to a scientist on another team about some obscure PhD topic from 10 years ago that somehow connected to [insert BD topic]. Spark. Connection. And then the BD people just sign off the paperwork. Hiring BD to do partnerships is like going indirect.

We offshored manufacturing to China and paid the price. Now we’re doing the same with biotech.  With clinical trials leaving the U.S., our leadership in medicine will follow. One policy change could reverse it. (Link to my OpEd in thread below)

Really interesting science by @leashbio, congrats to them, and everyone should be paying attention to the insights and issues they're discovering.

We just showed excellent small molecule binding prediction by a model trained *only* on data we collected ourselves experimentally, something we haven't seen anyone else do. blog post: leashbio.substack.com/p/we-now-know-… preprint: github.com/Leash-Labs/her…

This is why the @noetik_labs deal is so interesting. It’s defensible because they own the whole stack. It’s valuable because it tells you if the drug will actually work. This is the template for real value in biotech platforms generally, and especially AI/Tech Bio.

AI for small molecules is really hard, @leashbio is an example of how to do it right. A really innovative and daring data generation story. Must read and congrats to @allmeasures and @Andrewdblevins for building such a cool platform.

@RetroBio_ tech might not work, but if they're pitching, "I'm going to spend 1B to get through phase 3 and prove we delay aging", 5B might make sense if it does work. It's as good a thesis as @altos_labs or @Xaira_Thera or any number of other mega-rounds.

The best AI industry overview as usual from @nathanbenaich. Balanced on hype and where the industry is going, very exciting times ahead!

@SynBio1 It’s for a far more boring reason: it’s the easiest way to operate large organizations. And measuring the needed/correct amount of middle management is hard. I think every industry would say they have too many middle managers. No need for biotech specific causes.

Success in academia means becoming a PI and “leaving the bench” PhDs are trained in this culture. Actual lab work is for junior mentees. Important people don’t do that So naturally biologists in industry scramble to leave the bench and biotech is bloated with middle managers

There is a lot of market uncertainty in biotech/techbio today. Tech companies have historically navigated times like this by focusing on becoming default alive. I wrote a little blog post defining what that means and how we could apply it to bio as well: ankitg.me/blog/2025/08/2…

@pli_cachete Biotech. You basically get one clinical trial. It has to work or the company will likely die (or worse, patients might). Even when successful, often there’s not enough capital or time for the original startup to make improvements afterwards.

What are counter examples to this? Cases where it’s actually quite important to do it well the first time.

@draparente I do wonder if secondaries have happened in bio, have you heard of any?

Said the same to a friend the other day. This will be true for bio too.

@jrkelly @davidycli Except there’s a ceiling in biology. How many people are taking generic Lipitor as SoC, and to my knowledge nothing has come along that’s better. Software basically can always get better, drugs seem to cap out. (Yes various PCSK9 strategies might change that but still TBD)

@davidycli hahah truth, benchtop lab equipment def around that requires NT :p

What's really wild is the end markets for biotech are actually bigger than the markets for tech -- human health, food, longevity, building materials, etc, etc. We really need to improve our tech stack -- we've been largely frozen at the lab bench for the last 20 years.

Thrilled to see this come out. It was great working with Grant Koch and Scott Lokey, and congrats to Meghan Lawler, Kaileigh Cloutier-LeBlanc, Neil Carlson and LaShadric Grady on leading the collaboration. So many innovations in DELs waiting to be built! chemrxiv.org/engage/chemrxi…

AI today depends critically on being able to check the answer’s validity. I think UIs that make it easy to find and fix errors are key to realizing its promise.

@davidycli @ElliotHershberg @bwhite5290 Of course there will be successful services/software cos, but their value prop must be very obviously linked to creating more successful therapeutics companies. And their market cap will grow in direct proportion to the number of successful therapeutics they enable.

@davidycli @ElliotHershberg @bwhite5290 On the discovery side Schrodinger has a ~$1.5B market cap (also big % of TAM on the software side) where a company like Mirati was acquired for nearly $4.8B with a $1B bonus thrown in.

The unfortunate market reality for "Tech Bio" selling into pharma as a tech solution is dangerous work. The market is (usually) simply not large enough, and double whammy is that sales cycles are loooong let's take a look at the tech startups that have started in this space Formation Bio - clinical trial software that pivoted to tx drug discovery Vial - clinical trial infrastructure tech that pivoted to tx drug discovery Schrodinger - small mlc modeling software that pivoted to yep you guessed it - tx drug discovery many, many bioinformatics startups over past 5 years - few surviving many, many gene editing / CRISPR software startups a decade before that - few surviving running joke in this space is that you either die or live long enough to get into tx development the rare exceptions? Veeva - selling clinical trial management & commercial document management Benchling - electronic lab notebook and LIMS software the through line there - they sold into existing budget line items How do you invert this model? Thinking especially about the next crop of AI agents, LLMs, and other next gen tech You have to get out of the tech budget and get into services budget.