Labstep wants to fix the way science experiments are recorded and reproduced

Labstep wants to fix the way science experiments are recorded and reproduced

Labstep wants to fix the way science experiments are recorded and reproduced

Labstep, an app and online platform to help scientists record and reproduce experiments, has raised £1 million in new funding, including from existing investors. The company, whose team has a background in commercial R&D and academic research, including at Oxford University, is backed by Seedcamp and says it plans to use the new capital to double its team to 12, and for further product development.

This will include the launch of a marketplace for lab supplies, and is one of the ways Labstep plans to generate revenue. The startup will also add features to its app that streamline how scientists outsource elements of their research.

First conceived of in late 2013 and soft launched in 2015, Labstep has set out to digitise the lab experiment tracking and sharing process, and in turn give scientific research a major leg up.

As explained by CEO and co-founder Jake Schofield, science experiments are often recorded in an archaic way, relying on a mixture of pen and paper or entering resulting data into legacy software. Not only is this cumbersome but it also means that experiments are prone to mistakes and can be especially hard to replicate and therefore validate, either by a team working together internally or when sharing and cross-checking with the wider scientific and research community.

Enter: Labstep. The platform and app enables scientists to build libraries of experimental procedures — a bit like recipes — and then easily record progress when following a procedure in the lab, including building a timeline of the experiment. Procedures can also be shared with teams or more broadly, as well as deviated from in a transparent way. In fact, Schofield says one way to think about Labstep is as a ‘Github for lab experiments’. Procedures can be made public or private and can be optionally forked.

“Rather than following paper printouts, when actually carrying out your experimentation you can walk through these procedures step by step on a mobile device at the bench,” Schofield tells me. “Interactive features streamline and make it much easier to capture, comment, and record when you deviate from these processes”.

“Our API allows you to connect all the devices in your lab and automate the upload of results,” he explains. “Every action creates a timeline post, this automatic audit trail increases accuracy and saves the huge amounts of time normally spent writing a progress diary after the fact. You can form lab groups, like internal slack channels, that allow you to share these protocol libraries and real-time updates to see how your colleagues are progressing, this is massive as people are often collaborating and working in different geographical locations”.

In addition, the record of the steps that lead to a scientific conclusion can be attached to academic papers in the form of a URL so that other scientists can attempt to replicate the findings. This feature alone could go some way to tackling what the Labstep founder says is “a global reproducibility crisis,” estimated to cost billions per year in wasted research.

“At the point you publish your results, the competitive emphasis on keeping your research private shifts as you now want others to reproduce and validate your findings. We generate unique IDs that can be put in your publications and methods sections to link the protocols and the process that lead to these results,” he says.

As a route to monetisation, in the coming months Labstep will roll out a marketplace to make it easier to source the lab supplies needed to reproduce findings. It also plans to harness the real-time data that the Labstep app captures on how supplies in the lab are used, and Schofield says that by streamlining the ordering process, the reproducibility problem can be further addressed.

In another nod to collaboration, Labstep will also launch cloud features that allow users to outsource elements of the experimental process. I’m told that although outsourcing of research is commonly done in commercial R&D, it is used much less in academia.

Meanwhile, Labstep says it has users from over 600 universities globally including Stanford, Harvard and MIT in the U.S., and Oxford, University College London, Imperial College, King’s College and the Crick Institute in the U.K. It’s also not the only startup in this space to have got the attention of investors. Benchling, a graduate of Silicon Valley’s Y Combinator, raised a $14.5 million funding round a couple of weeks ago.

Labstep wants to fix the way science experiments are recorded and reproduced
Source: TechCrunch

Silexica, which optimises how disparate applications work together on autonomous cars, raises $18M

Silexica, which optimises how disparate applications work together on autonomous cars, raises M

As we inch closer to building increasingly autonomous cars, the complexities that these vehicles will present as programmable hardware are increasing, with some 250 applications ranging from cameras to navigation controls to weather sensors running on a typical high performance system today. Now, a startup that is building a system to help optimise that and let all of those vendors work together in a neutral way is announcing some funding to continue its growth. Silexica, a startup based out of Cologne, Germany, that has developed a set of tools to help map and optimise a wide range of applications across multicore processors — specifically the kinds of applications and computers that power self-driving cars — has raised $18 million.

The Series B is led by EQT Ventures, the newish firm that sits under the PE firm EQT Partners, based out of Stockholm. Existing investors Merus Capital, Paua Ventures, Seed Fonds Aachen (Silexica originally spun out of the University of Aachen) and DSA Invest all also participated in this round. The company has raised $28 million to date.

The funding, Silexica’s CEO Maximilian Odendahl (who co-founded the team with Johannes Emigholz and Weihua Sheng) told TechCrunch, will be used not just to increase its optimisation features — allowing for an increasing variety of services and functionalities to be monitored and diagnosed — but also to add a cloud component to how it monitors and processes information on a vehicle.

“Currently we are an on-prem solution, but we are building cloud platform,” he said. The company is also working on a simulator, so that multiple partners can work together on one platform to test their services and how they perform together. Its customers today include Denso, Toyota, Fujitsu and Huawei — underscoring the range of potential buyers of its tech.

The simulator points to one of the key reasons why companies like Silexica are emerging and attracting interest from the autonomous car industry. In large part, these systems are being built using components and technology from dozens or more vendors on top of the car company itself. But data, as people like to say, is the new oil, so what a vendor gathers from its specific sensors and services becomes valuable training information for better services. This means many of them are very guarded about what they share with others, and why, in turn, it is valuable to have an independent platform where that data mixes and is “seen” by no one else.

The other area where Silexica’s SLX tools are notable is that they work in real time to provide their diagnostics. Today, we’ve already seen some dreadful accidents involving autonomous vehicles not behaving in the way that we would have expected, resulting in fatalities. Inevitably, there will be more times that these systems don’t work the way that we think they will, and so any service that can improve how applications communicate and respond to each other will be increasingly essential — and in some cases fundamental — both to make the systems work better, and to make sure not just that vendor trust is in place, but that user trust is, too.

“In their quest to solve one of the largest challenges of the post-PC era, we believe Maximilian, Johannes and the rest of the team can steer Silexica into becoming one of the most important technology companies of this decade,” said Ted Persson, Design Partner and investment advisor to EQT Ventures who will be joining Silexica‘s Board.

There will be other approaches taken to solving this problem, too. Given that we are still at a relatively nascent stage of the race for self-driving vehicles –Odendahl estimates that fully-autonomous might not in use until 2025, and that is possibly optimistic — it will be interesting to see how this aspect of the stack plays out.

Updated with correction to date for autonomous car rollouts, which should have said 2025, not 20-25 years.

Silexica, which optimises how disparate applications work together on autonomous cars, raises M
Source: TechCrunch

Blockchain browser Brave starts opt-in testing of on-device ad targeting

Blockchain browser Brave starts opt-in testing of on-device ad targeting

Brave, an ad-blocking web browser with a blockchain-based twist, has started trials of ads that reward viewers for watching them — the next step in its ambitious push towards a consent-based, pro-privacy overhaul of online advertising.

Brave’s Basic Attention Token (BAT) is the underlying micropayments mechanism it’s using to fuel the model. The startup was founded in 2015 by former Mozilla CEO Brendan Eich, and had a hugely successful initial coin offering last year.

In a blog post announcing the opt-in trial yesterday, Brave says it’s started “voluntary testing” of the ad model before it scales up to additional user trials.

These first tests involve around 250 “pre-packaged ads” being shown to trial volunteers via a dedicated version of the Brave browser that’s both loaded with the ads and capable of tracking users’ browsing behavior.

The startup signed up Dow Jones Media Group as a partner for the trial-based ad content back in April.

People interested in joining these trials are being asked to contact its Early Access group — via community.brave.com.

Brave says the test is intended to analyze user interactions to generate test data for training its on-device machine learning algorithms. So while its ultimate goal for the BAT platform is to be able to deliver ads without eroding individual users’ privacy via this kind of invasive tracking, the test phase does involve “a detailed log” of browsing activity being sent to it.

Though Brave also specifies: “Brave will not share this information, and users can leave this test at any time by switching off this feature or using a regular version of Brave (which never logs user browsing data to any server).”

“Once we’re satisfied with the performance of the ad system, Brave ads will be shown directly in the browser in a private channel to users who consent to see them. When the Brave ad system becomes widely available, users will receive 70% of the gross ad revenue, while preserving their privacy,” it adds.

The key privacy-by-design shift Brave is working towards is moving ad targeting from a cloud-based ad exchange to the local device where users can control their own interactions with marketing content, and don’t have to give up personal data to a chain of opaque third parties (armed with hooks and data-sucking pipes) in order to do so.

Local device ad targeting will work by Brave pushing out ad catalogs (one per region and natural language) to available devices on a recurring basis.

“Downloading a catalog does not identify any user,” it writes. “As the user browses, Brave locally matches the best available ad from the catalog to display that ad at the appropriate time. Brave ads are opt-in and consent-based (disabled by default), and engineered to operate without leaking the user’s personal data from their device.”

It couches this approach as “a more efficient and direct opportunity to access user attention without the inherent liabilities and risks involved with large scale user data collection”.

Though there’s still a ways to go before Brave is in a position to prove out its claims — including several more testing phases.

Brave says it’s planning to run further studies later this month with a larger set of users that will focus on improving its user modeling — “to integrate specific usage of the browser, with the primary goal of understanding how behavior in the browser impacts when to deliver ads”.

“This will serve to strengthen existing modeling and data classification engines and to refine the system’s machine learning,” it adds.

After that it says it will start to expand user trials — “in a few months” — focusing testing on the impact of rewards in its user-centric ad system.

“Thousands of ads will be used in this phase, and users will be able to earn tokens for viewing and interacting with ads,” it says of that.

Brave’s initial goal is for users to be able to reward content producers via the utility BAT token stored in a payment wallet baked into the browser. The default distributes the tokens stored in a users’ wallet based on time spent on Brave-verified websites (though users can also make manual tips).

Though payments using BAT may also ultimately be able to do more.

Its roadmap envisages real ad revenue and donation flow fee revenue being generated via its system this year, and also anticipates BAT integration into “other apps based on open source & specs for greater ad buying leverage and publisher onboarding”.

Blockchain browser Brave starts opt-in testing of on-device ad targeting
Source: TechCrunch