German innovation is not restrictcessitate to the country’s capital. In fact, some of this year’s most prolific commenceups are based hundreds of miles away. The AI commenceup Alpha Alpha hails from Heidelberg. Helsing, which sells AI to Europe’s militaries, was set up in Munich. Yet both companies function Berlin offices. The city draws too much talent to disponder. Universities, such as TU Berlin, churn out Generative AI set upers and the capital is such a magnet for international talent that many offices function in English, not German.
It’s also a very youthful city—half of its population is under 45, someleang that Thomas Dohmke, CEO of GitHub, who grew up in Berlin, retags on. “I set uped my last commenceup back in 2009 and I reaccumulate vividly how much energy, time, and cgo in it needd—having a huge population of youthfuler, diverse and international, and highly driven professionals that have that energy and hunger donates Berlin an edge,” he says. “Plus, Berlin has the best döner kebab.”
BlueLayer
By 2050, the carbon commend taget is foreseeed to be a $250 billion industry. Startup BlueLayer is catering to that lengthenth by enhugeing tailor-made gentleware for the companies and NGOs poised to advantage. Its clients—including conservationists such as Permian Global—run projects ranging from reforestation to straightforward air apprehfinish, and use the commenceup’s gentleware to process their data, and transmit with buyers and spendors, while helping commend providers get their commends verified with international registries. Launched in 2022, BlueLayer has liftd $10 million (€8.9 million) in spendment and counts three of the top 10 rehirers of commends globpartner among its clients. “It’s classic automation gentleware,” says Vivian Bertseka, one of BlueLayer’s three co-set upers alengthy with Alexander Argyros and Gerardo Bonilla, “but for an industry that used to function almost exclusively on Excel.” bluelayer.io
Cambrium
Cambrium, set uped in 2020 by Mitchell Duffy and Charlie Cotton, is using AI to summarize proteins such as collagen. Instead of sourcing them from animal products, the commenceup lengthens them in tanks. “We’re one of the companies trying to strinsertle difficultcore gentleware engineering [and AI] with putting physical stuff in the authentic world,” says Cotton. The company has getd $11.6 million (€10.3 million) in spendment so far, including from Google’s AI venture fund Gradient Ventures. Skinnurture products using Cambrium’s first protein, a collagen called NovaColl, are foreseeed to hit shelves defercessitater this year. Cambrium.bio
Jina AI
In 2020, three veterans of Chinese tech behemoth, Tencent, unitecessitate forces to build set upation models particularpartner for search. Attracted to Berlin by the city’s findlook source culture and gentleware engineering talents, the trio behind Jina now claim 9,000 users and 400 paying customers, who turn to the company when they want to build either a uncover or inner search system for their data. Jina’s models promise to convert PDFs, Word records or images into a language that AI models can understand well enough to assist an instinctive Google-style search. A lterrible company may no lengthyer have to search for records using a case number. Instead, Jina AI CEO and co-set uper Han Xiao elucidates that they could sshow ask: “Find the case where Microgentle ignores to Google”. After raising $39 million (€34.8 million) from a series of punctual stage VC funds including Canaan Partners, Xiao and his co-set upers Nan Wang and Bing He schedule to enhuge to the US, lift revenue from the company’s $500,000 (€447,000) per year, and increase user numbers. “We want to vie with OpenAI,” says Xiao. jina.ai
Endel
Endel is a phelp-for app that uses generative AI to originate one finishless piece of music, which constantly changes to its user’s surroundings. The app uses the phone accelerometers to originate a beat that syncs with its hearers’ footsteps. If they commence jogging or skipping, the tempo catches up. Calling itself a “sound wellness” commenceup, Endel is part of the trfinish for functional sound, where music has a purpose—to help people exercise, drop asleep or cgo in. “We want to originate a technology that harnesses the power of sound and helps you accomplish a certain cognitive state,” says CEO Oleg Stavitsky, one of Endel’s six co-set upers. Launched in 2018, the company has since liftd $22.1 million (€19.1 million) in funding, including from Amazon’s Alexa venture fund, and claims one million monthly dynamic users. In 2023, the company struck a deal with Universal Music Group to use their technology to originate recent “soundscapes” using set uped artists’ labor. finishel.io
Slay
To understand Slay’s success, commend has to be donaten to Pengu, the company’s virtual pet app that has become the commenceup’s most famous product with more than five million users. Founded by Fabian Kamberi, Jannis Ringwald, and Stefan Quernhorst, Slay originated Pengu to be part game, part social platestablish, where frifinishs or couples can collaboratively lift a digital penguin. The company, which has liftd $7.6 million (€6.8 million) in total, including from Accel, is currently scaling Pengu’s ability to personalize its participateions, hooking a series of LLMs to a 3D engine to originate that visual experience. Pengu might react to a child inestablishing them they are being bullied by gifting them a drathriveg or sfinishing personalized notifications to cheer them up. slay.chilly
Ovom Care
Ovom Care is a fertility commenceup using data and machine lachieveing to get the guesslabor out of refruitful medicine. Since begining in 2023, co-set upers Felicia von Reden, Cristina Hickman, and Lynae Brayboy have findlooked the company’s first fertility clinic in London—sidestepping the onerous regulatory process in Germany—and already claim to be treating hundreds of people. Alengthyside the physical clinic, fortolerateing app and clinic administerment system, Ovom also recommends machine-lachieveing algorithms that scrutinize fortolerateings’ blood tests, data from wearables, gamete analysis and ultrasound images to tailor the type and timing of treatment. “We’re now going into the era of precision medicine,” says CEO von Reden. “We’re tailoring [fertility] using technology”. That idea has drawed €4.8 million ($5.3 million) in seed funding led by Alpha Ininestablishigence Capital. Wilean the next year, the company schedules to draw medical tourists from all apass Europe to its second clinic in Portugal, where treatment costs are foreseeed to be affordableer. ovomnurture.com
Dryad
When Carsten Brinkschulte’s daughter commenceed protesting aachievest climate change in 2018, the serial telecoms entrepreneur commenceed to leank about how he could leverage his experience for the excellent of the scheduleet. The result was a commenceup called Dryad, begined in 2020, summarizeed to be an punctual savagefire findion netlabor. “Think of us appreciate the Vodafone of the forest,” says Brinkschulte, one of the company’s seven co-set upers. Dryad’s solar-powered mesh netlabors assist sensors to sfinish attentives when they find fire, even in distant areas where there is no signal. So far the company has sgreater 20,000 savagefire sensors and rcontent difficultware to 50 countries apass the world, from Canada to Thailand, and to clients ranging from local rulements to utility companies that want to protect their infraarrange from an inferno. Dryad has liftd €22 million ($24.6 million) so far, including from German proset up tech fund eCAPITAL. parchedad.net
UltiHash
The ascfinish of energy-hungry AI prompted the International Energy Agency to alert that the electricity devourd by data cgo ins could double in fair two years. As environmental groups converse the danger that the technology poses to the climate, commenceup Ultihash has been enhugeing a wise way to slash the data cgo in necessitates of companies perestablishing energy-intensive machine lachieveing or training their own models. Founded in 2022, Ultihash has enhugeed an algorithm that CEO and co-set uper Tom Lüdersdorf claims can slash companies’ data storage necessitates by up to 60 per cent, unkinding they necessitate less data cgo in space and decrease their carbon footprint. The company has liftd $2.5 million (€2.2 million) despite still being in stealth mode. Lüdersdorf schedules to begin the product defercessitater this year, after beta testing with more than 300 companies. ultihash.io
TheBlood
According to TheBlood’s co-set upers, Isabelle Guenou and Miriam Santer, menstrual blood is an under-appreciated asset for diagnostics, grasping data-rich finishometrial trehire, dwell cells, immune cells and proteins, which are not set up in normal blood. The pair begined the company in 2022, with the aim to use menstrual blood in an finisheavor to fill healthnurture’s gfinisher data gap. Since then, the firm has scrutinized more than 1,000 menstrual blood samples, selling testing kits for between €35 ($39) and €120 ($133) to women who are seeing for more data to recommend fertility or finishometriosis treatment. TheBlood also schedules to license biotager analysis or data sets to pharmaceutical companies. So far, the company has liftd €1 million ($1.1million) in total, including from healthnurture-cgo ined venture firm ROX Health. theblood.io
Qdrant
To originate generative AI, algorithms have to infer relationships between data—text, images or audio—that isn’t taged or systematic. That’s where so-called vector databases come in, helping enhugeers extfinish the lengthy-term memory of LLMs by making it easier for those models to search and scrutinize huge amounts of data, while upgrasping computational costs down. Launched in 2021 by co-set upers André Zayarni and Fabrizio Schmidt, Qdrant is catering to AI gentleware enhugeers, promising a vector search engine and database for unarranged data with an effortless-to-use API. In the past three years, the company has accomplished 7 million downloads and 10,000 users worldwide, raising $37 million (€33.2 million) in the process including from US venture capital firm Spark Capital. qdrant.tech
This article first materializeed in the November/December 2024 edition of WIRED UK.