Zebeth Media Solutions

Startups

After 40 million app downloads, PhotoRoom raises $19 million • ZebethMedia

French startup PhotoRoom has raised a $19 million Series A funding round. The company develops a popular photo editing app for e-commerce vendors and small businesses. In particular, it helps you remove the background behind objects you are about to sell so that your photos look more professional. Balderton Capital is leading the Series A round with angels from Facebook, Hugging Face and Disney+ also participating. Existing investor Adjacent is also putting more money in the company. PhotoRoom isn’t the only app that helps you remove photo backgrounds. Another popular app in the “Graphics & Design” category on the App Store is Pixelcut. More generic apps, such as Picsart also have background removal features. But PhotoRoom has focused on one niche in particular — small businesses reselling objects on eBay, Poshmark or Depop. In just a few taps, these marketplace experts can process a batch of photos and create images that are ready to be used. It can help them save a lot of time and sell more products. Behind the scenes, PhotoRoom uses deep learning to identify different objects and elements on a photo. A flat photo becomes a multi-layer images, which means that users can delete the photo background, blur it or replace it with something else entirely. The same technology can be used to retouch image and remove unwanted objects. PhotoRoom has a large library of background templates and helps you export images in different formats with each platform constraints in mind. And it’s been working remarkably well as the company managed to attract 40 million downloads on iOS and Android. There are 7 million monthly active users and hundreds of thousands of users pay a subscription fee to unlock all the features in the app. It currently costs $9.99 a month or $69.99 a year. Up next, PhotoRoom wants to bring generative AI to its app, starting with Stable Diffusion. That’s why it is raising money as the company has been profitable with its current team. “Until now, we grabbed an object and we erased the background. But then users have been looking for a background to put the object back in front of a template,” co-founder and CEO Matthieu Rouif told me. With a text prompt, PhotoRoom will generate a marketing product photo based on your object. The feature will be available in beta for some users at first. It will be rolled out to all users progressively. “By focusing intently on user needs, Matthieu and Eliot have created a product that stands out from the rest. The importance of online photography is immense and PhotoRoom has both the traction and the ambition to become a market leader,” Balderton Capital partner Bernard Liautaud said in a statement. Image Credits: PhotoRoom

Twitter said to delay verification check mark rollout until after US midterm elections • ZebethMedia

Twitter is reportedly delaying the rollout of verification check marks to subscribers as the social network attempts to steer clear of possible impact to Tuesday’s midterm elections. The Elon Musk-owned social media company had planned to rollout the revamped version of its subscription service, Twitter Blue, on Monday. The firm started testing the new features on Saturday, according to an app update note. Twitter did not immediately respond to a request for comment. The reported move to delay the rollout of verification badges comes just days before the U.S. midterm elections, with polls closing on November 8. The New York Times reports that the move was aimed at limiting the potential fallout of verified users impersonating political or government figures, such as President Biden, or news outlets claiming false results that may discourage others from voting. Jen Easterly, the director of CISA, the U.S. government agency tasked with overseeing election security and protecting voting infrastructure, said last week that the agency it found “no information credible or specific about efforts to disrupt or compromise,” but warned of the ongoing risks that disinformation campaigns sown by foreign actors aimed at undermining confidence in the elections system. Security experts like Chris Krebs, the former CISA director who was fired by President Trump for refuting false claims of election fraud, warned that Twitter’s new verification rules would be a “major risk” ahead of the midterm elections. At the time of writing, several verified users were impersonating Elon Musk, Twitter’s new owner, despite the risk of having their accounts suspended for impersonation. The revamped Twitter Blue, which will cost $7.99 a month in the U.S., includes a range of new features including the coveted blue check mark to anyone who subscribes as well as cuts down the number of ads served to them on the platform. Musk is betting on turning the subscription service into a major revenue source for Twitter, which he acquired last month for $44 billion. Musk financed $13 billion for the purchase from banks and needs to pay more than $1 billion a year in interest payments. The company began cutting costs earlier this week, laying off roughly half its workforce, or about 3,700 jobs. In a series of tweets earlier this weekend, Musk claimed the company’s revamped Twitter Blue “can beat” the ad-revenue YouTube offers to creators, and that he was also working on fixing the search functionality on Twitter. Twitter will soon allow users to attach long-form texts to tweets, he said. Twitter plans to roll out the revamped Twitter Blue to subscribers in many markets in the coming months. In response to a tweet, Musk said he is hopeful that Twitter Blue will roll out to users in India “in less than a month.”

How ButcherBox bootstrapped to $600M in revenue • ZebethMedia

Some of the best companies only come about because they found a problem worth solving. For Mike Salguero, CEO and co-founder at ButcherBox, the problem and opportunity in the extraordinarily broken space of meat production and distribution simply could not be ignored. Armed with an idea for how to do things differently, the company ran a Kickstarter campaign back in 2015, which drew the attention of its first thousand customers. From there, the company has continued to grow. At the recent Creative Technologist conference organized by venture capital fund Baukunst, Salguero shared that the company has seen $600 million worth of revenue without taking a penny of external investment and talked about some of the lessons he learned along the way.  A rocky start ButcherBox isn’t Salguero’s first rodeo. His first company was CustomMade.com, which raised $30 million in venture capital from First Round Capital, Google and Atlas Ventures in a series of funding rounds. But in spite of all the money it raised, the company wasn’t successful. “My experience was really bad. We lost everyone’s money, which I felt a lot of shame about,” Salguero recalls. “At the very end, I had diluted myself so much, I owned just 5.5% of the company. The business failed, and we ended up going bankrupt, losing everyone’s money.” After that, Salguero decided to walk a very different path with his next company, which he started after being confronted with a very personal problem. His wife has a thyroid condition, and in the process of doing an elimination diet to figure out what foods she might be intolerant to, they learned about grass-fed beef. However, this kind of meat was hard to find in the supermarkets in Boston. “While CustomMade was falling apart, I started calling farmers and asking them if I could buy a half-share of meat,” Salguero laughs. That’s a lot of meat, and he describes it as “basically two trash bags full of beef.” “I was meeting meat farmers in parking lots, buying a couple of trash bags full of meat — I’m sure that didn’t seem sketchy at all,” he said. “But it was too much meat for my freezer, so I ended up selling the excess meat to friends or people I was working for.” Some of his buyers repeatedly told him that it would be much better if the meat was delivered to their houses, and thus, the basic idea for ButcherBox was born. Meat in the mail “I got obsessed with the idea and started researching how you ship meat in the mail. I had no idea how to do it. But I’m a big believer in finding people who have done something before and then asking them for help. It skips a lot of the hard work,” Salguero explains. “I found the former head of operations of Omaha Steaks, which at the time was the big behemoth of meat in the mail. And he just said ‘Oh, yeah, my non-compete just ended. I’ll be glad to help you.’ He put all the pieces together at the beginning.” Then everything started happening all at once. Salguero was fired from CustomMade and even though he had aspirations of taking a 100 days off, going on a silent meditation retreat and recharging, he threw himself into building ButcherBox less than a week later. He hired an intern and launched a Kickstarter campaign in September of 2015, a decision made out of a desperation to never raise money again. Fundraising wouldn’t be necessary, he thought, as he wanted to do this as a hobby rather than as a big business. “I’m only going to put in $10,000 into this thing,” Salguero recalls deciding, adding that he vowed to keep things light and easy. “I gave equity to the Omaha Steaks guy, and I gave equity to the branding studio, which in retrospect was a mistake, because I had way too low of a valuation.”  Mike Salguero, CEO at ButcherBox speaks at the Baukunst Creative Technologists conference. Image Credits: Haje Kamps / ZebethMedia All aboard the rocket ship “We agree with vegetarians.” Mike Salguero, CEO, ButcherBox The company had a goal of $25,000 for the crowdfunding campaign, but it ended up raising eight times that amount in preorders. It soon converted a lot of the preorder customers into subscribers, and the rest is history. The company went from revenue of $275,000 in 2015 to $5 million in 2016, then $31 million in 2017 and kept growing. When COVID-19 hit, the meat-packing industry didn’t fare well, but ButcherBox’s revenue just kept growing as people started subscribing to home delivery services like there was no tomorrow. In 2019, the company had revenues of $225 million, but the pandemic tailwinds nearly doubled its top line to $440 million. In 2021, the company recorded $550 million, and this year, Salguero is optimistic his company will go past the $600 million mark.  “This whole time, I’ve just been on a rocket ship,” Salguero says. Beyond the numbers, the company has continued to stay true to its original mission of trying to make a difference. ButcherBox became a certified B corp in January 2021, joining the ranks of other heart-forward companies such as Allbirds, Ben & Jerry’s, King Arthur Flour and Patagonia, and further fortifying its aspirations as a company that takes a stand. Growing without external investment Figuring out how you build and grow a company without external investment is an exercise in scrappiness, but Salguero’s team had a few tricks up its sleeve, starting with the Kickstarter campaign and a number of communities who cared deeply about how and what they eat. The company figured out how to growth-hack its way to success by tapping bloggers and nutritionists. “You said eat grass-fed beef,” the company would tell them and created an affiliate model to help incentivize them to promote its products. “We don’t have any money, so we can’t pay you up front, but we will pay you for every box that person

The fintech layoffs just keep on coming • ZebethMedia

Welcome to The Interchange! If you received this in your inbox, thank you for signing up and your vote of confidence. If you’re reading this as a post on our site, sign up here so you can receive it directly in the future. Every week, I’ll take a look at the hottest fintech news of the previous week. This will include everything from funding rounds to trends to an analysis of a particular space to hot takes on a particular company or phenomenon. There’s a lot of fintech news out there and it’s my job to stay on top of it — and make sense of it — so you can stay in the know. — Mary Ann Wow, I take off one week and come back to all hell breaking loose in the fintech world. Sadly, it felt like we got news of layoff after layoff. I’ll attempt to round up as many of them as I can here: Chime confirmed that it is letting go of 12% of its employees. This equals about 160 people. According to an internal memo obtained by ZebethMedia, Chime co-founder Chris Britt said that the move was one of many that would help the company thrive “regardless of market conditions.” In the memo, Britt said that he and co-founder Ryan King are recalibrating marketing spend, decreasing the number of contractors, adjusting workspace needs and renegotiating vendor contractors. Opendoor announced it was letting go of 18% of its staff. This is around 500 people. Opendoor co-founder and CEO Eric Wu said his company, a publicly traded real estate fintech, was navigating “one of the most challenging real estate markets in 40 years.” Chargebee has laid off about 10% of its staff. As reported by Jagmeet on November 2, “Chargebee, backed by marquee investors including Tiger Global and Sequoia Capital India, has laid off about 10% of its staff in a ‘reorganization’ effort due to ongoing global macroeconomic challenges and growing operational debt. The Chennai and San Francisco–headquartered startup, which offers billing, subscription, revenue and compliance management solutions, confirmed to ZebethMedia that the update impacted 142 employees.” Stripe lays off 14% of its staff. As reported by Paul, “Stripe has announced that it’s laying off 14% of its workers, impacting around 1,120 of the fintech giant’s 8,000 workforce.” In a memo published online, Stripe CEO Patrick Collison conveyed a familiar narrative in terms of the reasons behind the latest cutbacks: a major hiring spree spurred by the world’s pandemic-driven surge toward e-commerce, a significant growth period and then an economic downturn ridden with inflation, higher interest rates and other macroeconomic challenges. Danish startup Pleo may lay off 15% of its workers. Jeppe Rindom, co-founder and CEO of Pleo — which less than one year ago raised $200 million at a $4.7 billion valuation — revealed that the company’s new strategy will impact 15% of its roles. He added that “up to 150 of our colleagues may have to leave.” Pleo is a developer of expense management tools aimed at SMBs to let them issue company cards and better manage how employees spend money. Credit Karma, now a subsidiary of Intuit, has “decided to pause almost all hiring.” This is according to an internal email sent to employees by chief people officer Colleen McCreary. McCreary referenced “revenue challenges due to the uncertain environment.” This was reiterated in Intuit’s fourth quarter earnings call, during which the company shared on November 1 that “all Credit Karma verticals have been negatively impacted by macro uncertainty. Credit Karma experienced further deterioration in these verticals during the last few weeks of the first quarter.” Remote online notarization services provider Notarize cuts its team by 60 people. A spokesperson told me via email that “the reorganization impacted nearly all teams and the decision was in service to the larger strategy we have been enacting at Notarize, and will enable us to move faster to best serve our customers.” The spokesperson added that in September, one small real estate–focused team was laid off in response to both its strategy shift and “the drastic drop in demand from the specific customers that they served.” The recent layoffs follow a larger layoff in June that impacted 110 people. Prior to that reduction, Notarize had about 440 employees. It currently employs 250 people across the United States. I wrote this newsletter on November 3 because I’m leaving on a trip to celebrate my 20th wedding anniversary, so it’s possible that more layoffs took place between then and now. 🙁 What this means for the broader fintech world is not yet clear, but when well-funded companies such as Chime, Stripe and Pleo are cutting staff, it is no doubt sobering for all the players — small or large — in the space. Special thanks to TC senior reporter and very nice guy Kyle Wiggers for helping me draft the Weekly News and Fundings and M&A sections below so I could get offline and pack for my trip! Weekly News Jeeves, the fintech startup that recently raised $180 million at a $2.1 billion valuation, told ZebethMedia via email that it has launched a service called Jeeves Pay that it’s billing as a “credit-backed business payments solution” for enterprise customers. At a high level, Jeeves Pay lets customers use their existing credit line to send wires or pay vendors, ostensibly solving the problem of having to rely on cash or revenues to fund local and cross-border business and vendor payments. Jeeves Pay is available now to all Jeeves customers “where permitted by applicable local laws and regulations,” the company says. Brex sees startups as one of the key avenues to growth in the corporate card and spend management market. To that end, the company on Wednesday announced a partnership with Techstars to extend Brex services to companies within the accelerator, following similar tie-ups with Y Combinator and AngelList. For the duration of the accelerator, Techstars participants will get a Brex platform support team, access to exclusive Brex events and free use of Brex’s Pry financial forecasting

Elon guts Twitter, Google shutters Hangouts, and the tech layoffs continue • ZebethMedia

Hey, all — welcome back to Week in Review, the newsletter where we sum up the most read ZebethMedia stories from the past week. And oof, what a week it was. Want this newsletter in your inbox every Saturday? Sign up here. Signed up? Let’s just dive right in. most read Mass layoffs at Twitter: It was Elon’s first full week as the boss of Twitter post-$44 billion acquisition. Sweeping layoffs were said to be on the way — and, well, they’ve begun. After a painfully impersonal heads-up email went out Thursday evening, entire teams are waking up to find their access suddenly revoked. With reports suggesting layoffs could impact up to half the company, Twitter employees have reportedly taken to referring to the whole thing as “the snap” (à la Thanos). A class action lawsuit has already been filed alleging that Twitter isn’t following the proper legal processes here. Layoffs everywhere: Meanwhile, news of tech industry layoffs continues to pour in. Lyft let go of 13% of its workforce, Stripe cut 14%, Opendoor reduced its workforce by 18%, Chime parted ways with 12%, and more. Meanwhile, both Apple and Amazon have reportedly gone into hiring freezes. Google kills Hangouts: We knew it was coming, but this week Google put the final nail in Hangouts’ coffin, shutting down the chat-focused web app (the Hangouts Android/iOS apps were shuttered last year) in favor of Google Chat. Of course, given Google’s history with chat apps, I expect at least two more to be launched and/or shuttered by the time I finish this newsletter. Falcon Heavy returns to space: This week SpaceX launched its Falcon Heavy rocket for the first time since 2019, finally moving forward on a mission that had been delayed (“due to payload readiness issues”) since late 2020. Amazon expands its Music service: “The company said it will now offer Prime subscribers a full music catalog with 100 million songs, instead of the previously more limited selection of just 2 million songs,” writes Sarah, “and will make most of the top podcasts on its service available without ads.” audio roundup Whats up in TC podcast land this week? Here’s some of the highlights: The Equity crew chatted about the ever-evolving role of the venture capitalist, and our friend Melia Russell from Business Insider stopped by to fill us in on her recent story about how “investors are rewriting the playbooks when it comes to maternity leave policies at their firms.” Amanda joined Darrell on the TC Podcast to discuss Elon’s “questionable plans” to change up how identity verification works on Twitter The Chain Reaction team dive into the growing list of troubles that have developed for Bitcoin miners in the last few months. techcrunch+ Not a part of ZebethMedia+ yet? Here’s what TC+ members were reading most behind the paywall: Pilot’s CEO tears down their $60 million Series C deck: Published in early 2021, this one blew up for some reason this week! Just a few weeks after raising a big Series C, Pilot CEO Waseem Daher sat down with Lucas Matney to break down what worked about their pitch deck. The most common pitch deck mistakes: Speaking of pitch decks, TC’s resident pitch expert, Haje Jan Kamps, has a list of the mistakes he’s tired of seeing in decks, having reviewed thousands of them.

SaaS and alts • ZebethMedia

Welcome to The ZebethMedia Exchange, a weekly startups-and-markets newsletter. It’s inspired by the daily ZebethMedia+ column where it gets its name. Want it in your inbox every Saturday? Sign up here. As much as I like spotting new trends, it is just as important to get confirmation on previous predictions we made or heard. This week brought us some fodder in that regard, on two sectors that are pretty high on my radar: SaaS and alts. Let’s explore.  — Anna Shrinking SaaS multiples, hard times for IPOs Alex and I spent quite a bit of time this week diving into Battery Ventures’ “State of the OpenCloud 2022” report. It brought some forward-looking data to our attention — for instance, on cloud adoption — but also confirmed something impossible to ignore: That SaaS multiples — enterprise value compared to revenue projections — are shrinking. “The median forward multiple for SaaS companies has fallen from about 16x forward revenues to roughly 6x today,” Battery general partner Dharmesh Thakker told us. Multiples haven’t only shrunk, but they have also range-compressed, with fewer rewards for the fastest-growing companies compared to slower-growing ones. There are many factors at play, but the gist of it is that profitability seems to matter again to the markets. As a result of that, we’re seeing the revenge of some old rules. “Adjusted for growth,” Thakker said, “companies today that show efficient growth as implied by the Rule of 40 (i.e., companies with a growth rate + free cash flow margin greater than or equal to 40) are trading at a premium to those that are growing without regard to profitability.” Note that it’s not either growth or profitability: It has to be both, and the bar to please investors seems to be getting higher and higher. A more demanding market is a worrying picture for the many unicorns waiting to IPO, as well as for their peers who already went public but struggle to maintain their market cap. Let’s also spare a thought for Alex, who may not get his hands on another juicy S-1 before Q2 2023.

Jack Dorsey breaks his silence, owns “responsibility for why everyone is in this situation” at Twitter • ZebethMedia

Jack Dorsey, who stepped down as Twitter CEO less than one year ago, finally addressed the layoffs that impacted approximately 50% of the company he co-founded in 2006. The workforce reduction, led by Twitter’s new owner Elon Musk, impacted thousands of people – and key teams working on human rights, accessibility, AI ethics and curation. “Folks at Twitter past and present are strong and resilient,” Dorsey said on Twitter on Saturday morning. “They will always find a way no matter how difficult the moment. I realize many are angry with me. I own the responsibility for why everyone is in this situation: I grew the company size too quickly. I apologize for that.” Dorsey, who also stepped down from Twitter’s board five months ago, added that he’s “grateful for, and love, everyone who has ever worked at Twitter. I don’t expect that to be mutual in this moment…or ever…and I understand.” Folks at Twitter past and present are strong and resilient. They will always find a way no matter how difficult the moment. I realize many are angry with me. I own the responsibility for why everyone is in this situation: I grew the company size too quickly. I apologize for that. — jack (@jack) November 5, 2022 This is Dorsey’s first public comment since Musk took over the platform last week. In the past, Dorsey said that Musk is the “singular solution I trust.” Leaked documents from the Elon Musk v. Twitter trial give some insight into how Dorsey was thinking about the future of the social media company. Dorsey texted Musk that he left because Twitter needed to become a new platform – one that isn’t a company. “I believe it must be an open source protocol, funded by a foundation of sorts that doesn’t own the protocol, only advances it. A bit like what Signal has done. It can’t have an advertising model,” Dorsey texted Musk. Yesterday, impacted Twitter employees used the hashtag #LoveWhereYouWorked, a riff on the internal hashtag #LoveWhereYouWork, to thank each other, say goodbye and share personal news. As one former employee put it, the new hashtag is a “bittersweet phrase — not because I’m gone, but because it’s gone.” Despite Dorsey’s departure from his official roles at Twitter, his silence was noticed. Musk, meanwhile, addressed the layoffs on Friday evening. “Regarding Twitter’s reduction in force, unfortunately there is no choice when the company is losing over [$4 million a day],” Musk tweeted. “Everyone exited was offered 3 months of severance, which is 50% more than legally required.” Current and former Twitter employees can reach out to Natasha Mascarenhas at @nmasc_ or Signal, a secure messaging app, at (925) 271 0912.

Making DAOs accessible for normals on ZebethMedia Live • ZebethMedia

Want to start a DAO? It’s not that hard. Want to join a DAO? It’s even easier, but there are several steps to get connected. Some of those steps are daunting. That’s why Alex Taub started Upstream, which attempts to make starting and onboarding for a DAO much easier. Tune in to the next ZebethMedia Live to hear form Alex and investor Karin Klein (Bloomberg Beta) on Upstream’s novel approach to decentralized autonomous organizations. This ZebethMedia Live event opens on November 9 at 11:30 a.m. PDT/2:30 p.m. EDT with networking. The interview begins at 12:00 p.m. PDT followed by the TCL Pitch Practice at 12:30 p.m. PDT. Apply for TCL Pitch Practice by completing this application. If you haven’t joined us before on Grip — our TCL online platform — click here to register for free and gain access to all ZebethMedia Live events, including ZebethMedia Live, City Spotlight, Startup Pitch Practice, Networking and other ZebethMedia community events, with just one registration. Already part of the ZebethMedia Live on Grip community? Click this link to add this session to your agenda!

How to land investors who fund game-changing companies • ZebethMedia

A lot of problems worth solving aren’t ones that you can solve in a year or two or even 10. For founders and investors alike, such long timelines can seem daunting. But for Gene Berdichevsky, co-founder and CEO of battery tech startup Sila, hard tech problems are also some of the most tantalizing. “It’s always a good time to be a hard tech startup,” Berdichevsky said at ZebethMedia Disrupt. “One of the reasons is that the world doesn’t change just because it should. It changes because someone goes after something insanely hard and actually succeeds at it.” Such hard.tech startups run the gamut from advanced batteries like those made by Sila to nuclear fusion, quantum computing, automation and robotics. Any tech that has the potential for such broad impact also has a massive potential market, and that means a certain class of investors are willing to be in it for the long haul. “Hire people to do the technical stuff. Keep an eye on it, but then go learn the other pieces.” Gene Berdichevsky, co-founder and CEO, Sila “We look for real step-change, game-changing technologies that are going to benefit everyone and we think that will drive a huge [total addressable market],” said Milo Werner, a general partner at The Engine. When Berdichevsky founded Sila, he believed his company’s technology, a silicon-based anode that promises to improve lithium-ion battery energy density by 20%–40%, would be a significant enough advance that it would have no problem finding a market. What he didn’t expect was how long it would take. When Sila’s first product debuted inside the Whoop 4.0 wearable last year, the path to market had been twice as long as Berdichevsky had expected.

Stability AI backs effort to bring machine learning to biomed • ZebethMedia

Stability AI, the venture-backed startup behind the text-to-image AI system Stable Diffusion, is funding a wide-ranging effort to apply AI to the frontiers of biotech. Called OpenBioML, the endeavor’s first projects will focus on machine learning based approaches to DNA sequencing, protein folding, and computational biochemistry. The company’s founders describe OpenBioML as an “open research laboratory” — aims to explore the intersection of AI and biology in a setting where students, professionals and researchers can participate and collaborate, according to Stability AI CEO Emad Mostaque. “OpenBioML is one of the independent research communities that Stability supports,” Mostaque told ZebethMedia in an email interview. “Stability looks to develop and democratize AI, and through OpenBioML, we see an opportunity to advance the state of the art in sciences, health and medicine.” Given the controversy surrounding Stable Diffusion — Stability AI’s AI system that generates art from text descriptions, similar to OpenAI’s DALL-E 2 — one might be understandably wary of Stability AI’s first venture into health care. The startup has taken a laissez-faire approach to governance, allowing developers to use the system however they wish, including for celebrity deepfakes and pornography. Stability AI’s ethically questionable decisions to date aside, machine learning in medicine is a minefield. While the tech has been successfully applied to diagnose conditions like skin and eye diseases, among others, research has shown that algorithms can develop biases leading to worse care for some patients. An April 2021 study, for example, found that statistical models used to predict suicide risk in mental health patients performed well for white and Asian patients but poorly for Black patients. OpenBioML is starting with safer territory, wisely. Its first projects are: BioLM, which seeks to apply natural language processing (NLP) techniques to the fields of computational biology and chemistry DNA-Diffusion, which aims to develop AI that can generate DNA sequences from text prompts LibreFold, which looks to increase access to AI protein structure prediction systems similar to DeepMind’s AlphaFold 2 Each project is led by independent researchers, but Stability AI is providing support in the form of access to its AWS-hosted cluster of over 5,000 Nvidia A100 GPUs to train the AI systems. According to Niccolò Zanichelli, a computer science undergraduate at the University of Parma and one of the lead researchers at OpenBioML, this will be enough processing power and storage to eventually train up to ten different AlphaFold 2-like systems in parallel. “A lot of computational biology research already leads to open-source releases. However, much of it happens at the level of a single lab and is therefore usually constrained by insufficient computational resources,” Zanichelli told ZebethMedia via email. “We want to change this by encouraging large-scale collaborations and, thanks to the support of Stability AI, back those collaborations with resources that only the largest industrial laboratories have access to.” Generating DNA sequences Of OpenBioML’s ongoing projects, DNA-Diffusion — led by pathology professor Luca Pinello’s lab at the Massachusetts General Hospital & Harvard Medical School — is perhaps the most ambitious. The goal is to use generative AI systems to learn and apply the rules of “regulatory” sequences of DNA, or segments of nucleic acid molecules that influence the expression of specific genes within an organism. Many diseases and disorders are the result of misregulated genes, but science has yet to discover a reliable process for identifying — much less changing — these regulatory sequences. DNA-Diffusion proposes using a type of AI system known as a diffusion model to generate cell-type-specific regulatory DNA sequences. Diffusion models — which underpin image generators like Stable Diffusion and OpenAI’s DALL-E 2 — create new data (e.g. DNA sequences) by learning how to destroy and recover many existing samples of data. As they’re fed the samples, the models get better at recovering all the data they had previously destroyed to generate new works. Image Credits: Stability AI “Diffusion has seen widespread success in multimodal generative models, and it is now starting to be applied to computational biology, for example for the generation of novel protein structures,” Zanichelli said. “With DNA-Diffusion, we’re now exploring its application to genomic sequences.” If all goes according to plan, the DNA-Diffusion project will produce a diffusion model that can generate regulatory DNA sequences from text instructions like “A sequence that will activate a gene to its maximum expression level in cell type X” and “A sequence that activates a gene in liver and heart, but not in brain.” Such a model could also help interpret the components of regulatory sequences, Zanichelli says — improving the scientific community’s understanding of the role of regulatory sequences in different diseases. It’s worth noting that this is largely theoretical. While preliminary research on applying diffusion to protein folding seems promising, it’s very early days, Zanichelli admits — hence the push to involve the wider AI community. Predicting protein structures OpenBioML’s LibreFold, while smaller in scope, is more likely to bear immediate fruit. The project seeks to arrive at a better understanding of machine learning systems that predict protein structures in addition to ways to improve them. As my colleague Devin Coldewey covered in his piece about DeepMind’s work on AlphaFold 2, AI systems that accurately predict protein shape are relatively new on the scene but transformative in terms of their potential. Proteins comprise sequences of amino acids that fold into shapes to accomplish different tasks within living organisms. The process of determining what shape an acids sequence will create was once an arduous, error-prone undertaking. AI systems like AlphaFold 2 changed that; thanks to them, over 98% of protein structures in the human body are known to science today, as well as hundreds of thousands of other structures in organisms like E. coli and yeast. Few groups have the engineering expertise and resources necessary to develop this kind of AI, though. DeepMind spent days training AlphaFold 2 on tensor processing units (TPUs), Google’s costly AI accelerator hardware. And acid sequence training data sets are often proprietary or released under non-commercial licenses. Proteins folding

Subscribe to Zebeth Media Solutions

You may contact us by filling in this form any time you need professional support or have any questions. You can also fill in the form to leave your comments or feedback.

We respect your privacy.
business and solar energy