2. Biotech with Demi - Antroph, Biosimulant and the journey to In Omni

Added: 2026-05-04

[00:00:00]In vivo, in vitro, and in silico, which are the three big Latin derived terms that describe where science experiments take place, specifically for biology research.

[00:00:10]In vivo means within the living, which talks about experiments performed within a living being, that's where you have the ones like lab rats, within an animal or a plant.

[00:00:20]In vitro talks about experiments performed in a glass, within a lab, you see your scientists putting prepared test tubes, those fancy stuff. So, vitrus Latin for glass.

[00:00:34]And in silico, which are experiments performed on models on a computer.

[00:00:40]Silico is not really a Latin word, it's kind of like a play on a mashup because in vivo and in vitro existed before. Then this Latin guy Pedro in 1999 came up with in silico, which is just to say silicon, silicon chips, silicon chips are the foundation for computing. So, in silico means the experiments you run in a computer.

[00:01:00]It is in softwares. So, all these three hypotheses that within 15 years they would all collapse into one.

[00:01:10]A world where the boundary between biological substrates and computational substrates dissolves.

[00:01:17]And where a simulation isn't just modeling biology, the simulation itself becomes biology.

[00:01:23]Swapping a kidney for a synthetic equivalent, or a neuron, a brain neuron for a silicon chip, becomes like becomes routine.

[00:01:32]And even total brain simulation where I can copy myself into a system and still be myself, right? I can upload my consciousness, or not maybe not consciousness, but I can upload myself into another system. Another system will have all my memory and another experience, and I can decode it and encode and decode back and forth. So, I believe the hypothesis that within 15 years, I mean not in 2040s, we would not have achieved everything, but would have had would have made significant progress where in vivo, in silico, and in vitro becomes one.

[00:02:02]And for that I coined the term in omni.

[00:02:06]My friends think it's a shitty term, but let history remember that it was Tim like that coined the term in omni, which is just that world or that system where experiment performed in a wet lab, in a laboratory with test tube and petri dish, and experiment performed on a computer, and experiment performed within a human being are indistinguishable because we built a unified system, a unified framework that that that allows them to be almost replaceable. So, I could replace my arm with a mechanical arm, with a bionic arm easily. I could simulate my heart my my heartbeat, or my kidney function, and just replace it easily, right? So, in omni, let history just remember that it's Tim like that formed the term. Just put it there, shameless plug.

[00:02:59]Yeah. And and about four months ago I released a video where I mentioned that disability should not exist and would not exist in the future.

[00:03:11]And I also kind of talked about two projects I've been working on, Antrov Robotics and Biosimulants.

[00:03:16]Antrov is a robotic project that just asks what happens when you try to emulate biology by embodying artificial intelligence into the physical world.

[00:03:25]Biosimulants asks what does it look like when you simulate everything biology in a virtual environment.

[00:03:31]So, I've made some progress on both of them, and I'll give an update for this.

[00:03:35]They are still very much in the MVP stage. [clears throat] And I'm doing this because it's going to help me with recruiting, with partnerships. So, if you think you want to partner or engage in any way, please feel free to reach out to me on any social media platforms.

[00:03:50]For Antrov, phase one is this one.

[00:03:54]I was just getting started, barest minimum. We're still fine-tuning the idea, we call it Goldie. Phase two was building on top of that for a bigger multi-agent system, which we call Blackie.

[00:04:05]And phase three focused on building more of the back end, I've been on a mobile app that runs on the biointelligence system.

[00:04:11]So, phase four, we are really we're now getting into building the embodied AI traffic itself that you can buy in your house. It could be a smart lab, it could be a table top robot.

[00:04:21]That's for Antrov. And then Biosimulants, which is still very much in the MVP stage. Basically, Biosimulants allows you to simulate everything biology.

[00:04:31]So, with Biosimulants we allow you to just plug in mechanistic models, AI models, and just simulate everything in a workflow, complex workflow. Here's a quick demo.

[00:04:40]I'll show I'll show you a quick demo of it in a minute for you to just see what it is and how it works, what it looks like.

[00:04:45]So, I have it open. Let me just show you a quick run of it in action.

[00:04:51]>> [music] >> I can create a lab or import, or use an existing lab I've run. Started the simulation. [music] As you can see it here, this is a model.

[00:05:00]I can see the parameters, there are different samples, the cycling steps, sampling steps. This is going to be a model that you give it a protein, you give it a ligand, and it tells you if they can bond, and gives you some probability of how well they bond.

[00:05:14]So, this has been run already.

[00:05:16]These are some results that were generated. And then you just click on the results. So, these are some results of some past run. This is a protein 3D structure, and within it you can see the ligand that was attached to it. Somebody just see the affinity prediction, the binding probability. And this was all generated by just running this model.

[00:05:37]Usually you'd have had to do a lot of coding to generate something like this, or download the model and somehow run it. But with Biosimulants, any form of model, whether mechanistic model or AI model, it's just available for you to just run.

[00:05:48]AlphaFold is a little bit proprietary, so it was not made available, but we've made over 4,000 models available already, and more will be made available. If I this is a lab, I can add more models to the environment by just clicking on add, I get a list of models that I can just select from and add to the environment. But they download it and make it available here.

[00:06:12]So, the good part of it is that you're not just training models together, they're actually an agent here that you can now use, can ask questions, and can now use to actually run a lot of your simulations yourself, a lot of experiments. For example, if I type what can you do?

[00:06:28]So, it tells you it can help you in a few ways, inspect current lab summaries, select a node. And you can just do some things, you can tell it what was the result of the last lab lab. And you can also tell it to run the simulations for you against a particular value. And this is particularly useful in terms like drug discovery, where a big amount of work is in repeating the cycles and trying [music] to check a lot of drug against something you're seeing, seeing if the protein binds with the drug. If it doesn't bind, you try another protein molecules. And [music] it just gives you a lot of experiments. This one, we have the agent here, everything just happens for you automatically.

[00:07:05]So, you can see the last run completed successfully on remote. Here's the pulse affinity, the working direction is to use AI to simulate biology end-to-end, and then use the simulation capability to build systems that augment and eventually replace biological substrates.

[00:07:28]Right? And pretty much at a high level.

[00:07:30]This is why I'm talking about the two projects.

[00:07:33]Like I said, this hopefully helps with some of the recruiting and partnerships.

[00:07:36]So, this is why I'm talking about the two projects.

[00:07:39]So, that's that's it. That's what I've been up to in the past 12, 15 weeks.

[00:07:46]When there is an update I'll show you more, and I'll be doing a lot more videos on Biosimulants where I'll actually just show certain models simulated in action and see how it comes out. Thank you.