Amazon and Google gave us smart homes, where household items are reordered automatically as soon as stock runs low, and all of our devices and apps are interconnected and centralised.
Letting the tech giants constantly monitor almost every aspect of our home lives might be a bad idea, but one startup has identified a totally different context where constant tracking is required, rather than tolerated: science labs.
Labstep is bringing the Internet of Things to lab researchers, and they’re giving the tools away for free to academic institutions, subsidised by charging commercial biotech labs for the same product.
It may seem like a strange choice of business model for a young startup, but the Labstep team of ex-scientists is on a mission to solve the biggest problem in the scientific community: the reproducibility crisis.
Experimental science is in a “reproducibility crisis”
We trust the results of scientific experiments. We assume that these results are not one-off flukes and that repeating the experiment would prove this, by producing the same results.
In the last few years, scientific researchers have very publicly aired their concerns that this is not the case, that important results cannot be reproduced.
The problem isn’t new, but now it’s under the microscope. Almost three-quarters of scientists who have tested another researcher’s experiment failed to achieve the same results. The majority of researchers (90%) believe experimental science is in a “reproducibility crisis”.
Yet the digital revolution has barely penetrated science labs.
Labstep’s two founders, biochemist Jan Domański and biomedical researcher Jake Schofield, have seen this first hand. In both university labs and biotech companies, researchers are using printed instructions, notes are taken using pen and paper, and experimental data is collected and stored in hundreds of files spread across the hard drives in each different piece of equipment.
“Knowledge transfer is horrendous,” says Schofield. “If a researcher leaves, an experiment goes into a black hole because nobody knows how to carry on doing that work. When an experiment is recorded in a paper notebook, that information can never be shared and searched as you could do online. It is effectively lost.”
So far, innovation has simply meant using electronic notebooks instead of paper ones. But this doesn’t make work easier, more efficient or less error-prone; often it increases the workload for overburdened scientists, rather than replacing manual tasks with something better.
In a perfect world, science would be, well, an exact science, without making extra work for the scientists.
Labstep gathers all the research data in one accessible place
Labstep does not design smart instruments or robotic hardware for lab automation. A host of other companies are already established in this space, like Opentrons making affordable robotic pipette machines, Finnish company Spectral Engines selling bluetooth-connected material-identification scanners, and TetraScience, which provides hardware sensors to monitor temperature, power, humidity and open/closed doors for incubators and refrigerators.
Even the long-established equipment manufacturers are shifting from selling individual instruments to making robotic equipment modules to be used in a “build-your-own” automated lab workstation. The next generation of science labs will be cloud-based, with devices in centralised warehouses controlled over the internet, and no human in sight.
But automation on its own doesn’t help researchers easily generate a clear step-by-step record of exactly what happened and all the experimental conditions. The Labstep team believes this is the bigger opportunity, and the key to solving the reproducibility crisis.
“We're not a hardware company,” says Schofield. “There's a lot of very cool people making connectable devices. But the problem is actually just that there's not a platform where you can coherently tie everything together. We're trying to be that place.”
It achieves this by turning experiment protocols, which are usually printed out, into interactive instructions in an app, so researchers can tick off each step as they go. Experimental devices are connected to the platform so that data is automatically captured and saved in the right place.
Researchers can take notes and photos and save them under specific steps of the experiment. They can add comments to work done by co-researchers anywhere in the world, share results with collaborators, and publish the results alongside “metadata” about their experimental process.
The platform is designed for human researchers, but it is just as useful for entirely automated cloud-based experiments. Already it easily enables researchers to design, share and collaborate on experimental design, and run the experiment remotely on any internet-enabled device.
We’re very lucky that Labstep adoption has spread virally and organically. Once we get a colony of users at a key institution, they promote Labstep for us,
In time Labstep will use its own data and publication databases to help researchers find experts who have done similar experiments so that they can collaborate on projects, read relevant work or give each other advice.
Even though “research tech” hasn’t yet secured its place in the tech sector’s vocabulary, Labstep has shown that others also believe in the opportunity. It has raised £1.8m over two seed rounds, and graduated from Google’s startup residency programme in London.
The founders’ connections in commercial biotech and academic research labs have led to a number of distribution partnerships, including London’s biomedical research centre, The Francis Crick Institute, and The Wellcome Sanger Institute, specialising in genetics research in Cambridge. One of Europe’s largest science-funding bodies, the Foundation for Polish Science, is recommending Labstep is used in every research project it supports.
Creating a unified comparative marketplace
Schofield’s go-to analogy is that Labstep is like a digital, interactive recipe book for science experiments, with uploaded “recipes” (experimental protocols) converted into interactive instructions which automatically produce a report of exactly what was done to create the end result.
He says: “If we’re talking to a researcher, it's easy to explain what Labstep does using their specific experiment example. But talking to investors we have to take it out of context and say it's like baking: you can design your recipes, then you can walk through those steps in the kitchen, then you can take a photo of that brownie. And then you have the results.”
Labstep creates some of these supplementary tools itself, including functionality to create and share shopping lists of supplies the lab needs to buy, and keep track of how much has been used as the experiment runs. Schofield says that tracking exactly what was used in an experiment is just as important for reproducibility as tracking what was done. Different batches of the same antibody could produce different results, for example.
“Procurement is an absolute nightmare,” says Schofield. “You start an experiment only to realise that somebody has already finished that reagent, or you have no idea what reagent to purchase.
“There's currently no unified comparative marketplace, so we want to be the recipe book that then makes it easy for people to order their ingredients.”
Eventually, Labstep’s data will show users which products are the most popular, or which are recommended for experiments like the ones they are running, just like on Amazon. If it takes off, Labstep’s referrals to suppliers could be a vital revenue source.
Labstep seems to have boxed itself into a corner with its business model, by committing to providing the platform for free to academic institutions, and also promising that they will never monetise user data by selling to advertisers or other interested third-parties.
Its challenge is if revenue from commercial biotech companies (who will have to pay a licence fee) can support all the free users and make a profit for the startup.
Schofield is confident that Labstep can attract a sufficient number of paying customers onto the platform. Its users are up 80% this year, and he hopes that the research community’s enthusiastic adoption will encourage for-profit biotechs to also get on board.
“We’re very lucky that Labstep adoption has spread virally and organically,” he says. “Once we get a colony of users at a key institution, they promote Labstep for us because they collaborate with other institutions, including biotech companies.”
“All we need to do is target the key scientific clusters.”
