BioBright

CHARLES FRACCHIA: So I'm Charles Fracchia. I'm the CEO and founder of BioBright. BioBright is a small biotech company whose goal is to augment the human ability of doing science in the lab. By that, I mean we create voice assistant augmented reality systems that actually augment the human capability to do science, to collect information, centralize it, and analyze it.

So the typical lab environment today is it is an environment that is actually very low, technologically speaking. It is very old-fashioned. And many of the tools that we have are decades old and actually are not conducive to doing the data driven discovery that should be happening really in this field.

You have the tools like bipeds. You have equipment that doesn't connect to machines or has outdated connection mechanisms that have proprietary formats. All of that stands in the way of doing longitudinal and data driven discovery in this environment. So this is one of the major areas where we are changing the game of how biology is done in the lab is we have technology that allows you to actually collect the information when it's generated when and where it's generated, centralize it, and again analyze it in an enhanced way.

There's a major issue right now in the biotechnology biomedical research field, which really is the reproducibility issue. It's been dubbed the reproducibility crisis, and it's been actually amounted to about between $10 and $50 billion annually. And that's in the US alone. This is due to the fact that the tools really are not, again, conducive to this ability to reproduce these experiments.

So more often than not, when a scientist does an experiment, whether it succeeds or it fails, it's very difficult to actually determine the root cause of that success or failure. The tools really aren't collecting of the information that is necessary to determine that, so you will have problems in the academic world where papers are not really reproducible.

We've seen this recently in a number of high profile cases, not because it's malicious, in most cases, just because the information that needs to be recorded is not recorded.

The information in biological systems runs everywhere from picoseconds to hours, days, months. Humans are quite good at collecting information in kind of the minute range, but anywhere around that is very difficult-- surely, very difficult for a human to collect information picosecond range. But also it's very difficult for the human to draw insights from hours or even days or numbers of experiments, which may have been done across months sometimes. So really, this is the root cause of this reproducibility crisis is that there's a fundamental difficulty in the science, but the tools are inadequate to actually solve this problem, collect that information.

CHARLES FRACCHIA: So the BioBright interface is built to hook into sensors-- some that we've made ourselves, some that are commercially available, as well as by equipment that is routinely available in the lab-- and collect that information across [? scales ?] automatically, without the scientists needing to do an active action for it. That means that the information is pervasively collected, centralized in one place.

And now the scientists can spend their time actually trying to uncover the root cause of success or failure, but really stay in the experiment. They don't have to take their mind off of the experiment to say, oh, I now need to go collect this piece of information every 15 minutes. This is a very tedious environment right now. So we do all the design, all the hardware, and all the software work necessary to enable this centralized collection.

Today, in the laboratory, a scientist would generate data either by actually pipetting liquids around or creating vessels that contain this material, then typically go to a machine. This machine would either spit out a piece of information immediately or later on. The scientist is responsible for coming back to that machine or centralizing this information right now. So this is one of the major issues that we see.

The other is also that the moment between data generation and data collection is often actually quite far apart. So when a scientist goes in the lab and does an experiment, they are generating data. Their hands are often busy because they are generating, using tools, etc. They also have gloved hands that may be contaminated, right?

So one of the major issues that you have here is that because of the scapula, because of this immersion that the scientists really need to immerse themselves in doing the protocol, they put off, necessarily, the time to record this information. And so it ends up being recorded at the end of the day. A lot of information gets lost that way. Small deviations from the protocol are often not recorded, and so that leads to a wealth of information that just is missing.

So for example, if we were to take an example of a sample that may be a little bit more vicious than usual, this is an observation that the scientists may do at this point in time. They might write it down on a Post-it. If they're really disciplined, they might write it down in a notebook, but there it stays.

It's very difficult to draw that out of the context, and then look at the longitudinal aspect of the experiment. That scientist may have done the experiment five times already, where the viscosity has suddenly changed, but it's very, very difficult. It's nearly impossible, today, to actually see that information longitudinally.

CHARLES FRACCHIA: BioBright is a platform that unifies a number of data streams. We centralize those data streams. In cases where we have needed to create our own hardware to go and collect the data that was otherwise not collected, we have done so. An example of that is with a camera system that we have. We have a dual camera system that is both visible light and infrared, which allows us to do thermal imaging-- very important for biomedical experiments. Then we have the systems that allow you to collect information that was not otherwise capable.

You can place these cameras across your laboratory where necessary-- over benches, over specific stations, or even looking at specific areas of a robot if you're doing semi-automation approaches. And our interface allows you to seamlessly collect that information from different parts. Another part that we have is the hot folder system that allows you to interface with equipment directly. And so the computer that is running the equipment will be running our system, and that will automatically grab the data as it gets generated from that equipment.

That allows us to get virtually any equipment, piece of equipment, have this data automatically imported and centralized alongside the rest of our data streams. In other cases, we've also created small sensors, sensors that are actually fit for the purpose of doing biological research in the lab-- in particular, things that fit in vessels and tubes that are standard sizes.

BioBright's sensor network is a part of the BioBright interface. It allows you to collect information that was not previously collectible. We've designed the system with a lot of security in mind. So we've actually encrypted all the way down from the sensor all the way up to our system uniquely and in transit, which is something that's very important for data security.

Because of that, we've created a network that is self-healing, relies on a mesh network, encrypted, and that allows you to bridge the local environment of your experiment over to the global information that you may have gotten from the rest of the BioBright ecosystem-- like, for example, the hot folder, which allows you to import the data from the equipment, or again, the voice notes and the Darwin system that we use, the Darwin voice integration system that we use.

The way that looks in the lab using the BioBright system means that you can go around your lab just the way you did before. We are very focused on making sure that the scientists change as little as possible of their workflow during the day. You add a little microphone to your lapel. You are now able to interact with the BioBright system using a natural language.

And so our Darwin voice assistant allows you to collect voice notes, correlate data, and actually actuate the BioBright ecosystem to both collect information but also pull information back out. You can, for example, say, take a picture. It will automatically connect to our camera system, take even a thermal imaging picture. And that is automatically collected. That's it. You don't need to do any more work than that.

So in a sense, whereas before in the laboratory, scientists would go in, and we'd have to write everything down, our system allows your lab notebook to write itself, in a sense. It's an automatic system that collects the information, centralizes it for you, for the scientist, allowing the scientist to really focus on what matters, on what they've been trained to do-- analyze and understand the underlying data, not spend hours upon hours just collecting the piece of information with a lot of error-prone workflows.

CHARLES FRACCHIA: Yeah, the way we think about the biological research workflow, in recent years there's been a lot of interest in the automation. In fact, I was actually one of the early members of another MIT company named, Ginkgo Bioworks, that has built much of the early tools for automation in molecular biology. And I built a lot of the automation systems then. So I built a lot of knowledge understanding what automation was good for and what it was bad for.

And I think that is a lot of interest in automation and there should be. However, it's not the whole picture. One thing that we've-- one trend that we've seen, with the automation, is that it's mostly driven by computer scientists who have the tools, but who don't necessarily know the intimate problems of being in the lab for long hours of a time. They're not biologists.

And so BioBright inverted that. We are biologists who've learned computer science and iconic engineer so we could solve those problems. We're very intimate knowledge of the workflow. And that's why I talk about the necessity for human augmentation, putting the human at the center of the loop not taking them out of the loop like automation does. But those two things go hand in hand.

Automation is excellent at doing workflows and protocols 1,000 times over, but it is very brittle. It has no ability to adapt on the fly. It has no resilience to changes.

Research really, for the most part, is an exploration. Which means, you are changing parameters on the fly. You are adapting. You are deciding to ignore particular changes. And research often is at the beginning before development, before scale-up, even in pharmaceutical environments.

Research always happens first. And we found that most of the time and most of the reproducibility problems actually come from translating that workflow from one person to another, from one group to another, from even a geographical location to another and there are very few handles on that. Automation will not help with that.

Human augmentation on the other hand, allows you to provide almost superpowers to the scientists being able to, at a glance or with a voice command say, Darwin show me the average temperature that I have done this step at for the last three months. Or, show me how Mike did it last week. And pop a video.

So the way the BioBright platform works is that we centralize the information onto our systems. And then you can authorize specific users either in your lab or across your organization to highlight and review the information that you've done. So if I were to do a protocol say here in Cambridge, Massachusetts and it's the dead of winter, it may be very different from doing the experiment across organizations, say in San Diego, California, at the same time of year.

It may be very important-- it is very important for the scientist at the other end who is going to take that experiment to be able to see that information. That happens through our platform, and we are exploring ways of how to make that as easy as possible, potentially even with just an attachment to an email, et cetera. The access to the platform allows you to see all of that with the permissions, obviously, that you require.

But there are other ways that we are looking into two augment the relationship so that people don't have to change your workflow. This is really key. We don't want people to change their workflow in the way they share their experiments.