Dresden-based industrial IoT startup Packwise has secured an injection of seven-figure growth capital to fuel its international expansion and push the limits of its container monitoring platform.

The funding round saw the continued support of existing backers TGFS Technologiegründerfonds Sachsen and Hüttenes³, while bringing new investor SBG – Sächsische Beteiligungsgesellschaft on board through the RegioInnoGrowth (RIG) programme.

“This financing is a decisive milestone for Packwise. It enables us to consolidate our market position in Europe while significantly expanding our presence in North America. We are delighted with the confidence shown in us by our existing investors – and particularly pleased about the new investment by SBG as part of the RegioInnoGrowth program (RIG), which specifically supports growth-oriented technology companies such as Packwise,” says Gesche Weger, Managing Director of Packwise GmbH.

Founded in 2017 in Dresden, Packwise has carved out a niche in the digitalisation of intermediate bulk container (IBC) logistics. Its solution hinges on the Packwise Smart Cap, a plug-and-play IoT sensor that collects real-time data on location, fill levels and environmental conditions of industrial containers.

Paired with the company’s Packwise Flow software platform, the system forms a digital twin of supply chain activity, enabling better oversight, predictive analytics, and more sustainable container management practices.

The new funding will be directed towards three strategic pillars:

• Further development of the Packwise Flow platform
• The launch of new sensor variants tailored to specific container types such as steel drums and ISO tanks
• A substantial expansion of the company’s Houston-based US office.

The goal is to strengthen its presence in North America while consolidating its market share across Europe, where it already counts BASF, Merck, Solvay and Hoover among its 50-plus clients.

“With our participation in the ‘Innovation capital Saxony’ initiative, we want to specifically strengthen forward-looking business models such as Packwise’s, support them in their transition to the scaling phase, and further promote Saxony as a location for innovation. We are very pleased to be supporting a company whose technology has the potential to set new standards in industrial logistics,” adds Frank Tappert, Managing Director of SBG.

Long-time backer TGFS, which has supported Packwise since 2018, sees the startup’s progress as a textbook case of Saxony-born innovation scaling on a global stage.

“Packwise is an outstanding example of a growth-oriented technology company from the region that impresses with its scalable business model, technological excellence, and international appeal. By combining funds from existing investors such as TGFS and Hüttenes³ with RIG funds, we can invest specifically in growth-promoting measures and thus sustainably increase competitiveness,” adds Sören Schuster, Managing Director of TGFS.

As Packwise charts its next phase of expansion, the funding round solidifies its ambition to redefine container logistics through data, automation, and actionable insight — all underpinned by a robust industrial IoT stack.

The post Germany’s Packwise pockets growth capital for transatlantic push and next-gen container tech appeared first on EU-Startups.

Milan-based startup Maxi Mobility, which provides an electric fleet-as-a-service solution for professional drivers, has closed a €1.2 million funding round to expand its operations and accelerate the adoption of zero-emission mobility in Italy’s urban transport sector.

The round saw participation from UniCredit Bank and the Motor Valley Accelerator, part of CDP Venture Capital’s National Accelerator Network. Other co-investors include Plug and Play Tech Center, Fondazione di Modena, a group of angel investors, and Maxi Mobility’s own founding team.

“The arrival of a partner such as Motor Valley Accelerator represents a strategic validation of the work we have done and gives us the impetus to bring our vision to an industrial scale,” says Gian Paolo Incutti, CEO and Founder of Maxi Mobility. “Electric mobility can become truly accessible if it is built around the concrete needs of those who experience it every day.”

Founded in 2021, Maxi Mobility offers an all-inclusive subscription model designed for taxi drivers and corporate urban fleets. The service bundles electric vehicle leasing with unlimited charging, fleet management software, and 24/7 driver support – aiming to remove the main friction points that typically hinder electric vehicle adoption, such as upfront costs, charging complexity, and operational inefficiencies.

The newly secured funding will be used to scale up Maxi Mobility’s existing operations in Milan and Rome, further develop its proprietary digital platform, and enhance its charging infrastructure across strategic urban locations.

The company also aims to begin expanding into selected European markets where demand for electrified professional fleets is growing, supported in part by local environmental regulations and governmental incentives.

Maxi Mobility’s market focus is strategically timed. According to the company, there are currently over 13,000 active taxi licences across Milan and Rome alone, a number that is expected to grow with the 2023 Asset Decree, which encourages municipalities to issue new licences specifically for eco-friendly vehicles.

With a tailored model for these operators, the company is positioning itself as a key enabler of the electric transition in cities under regulatory pressure to cut emissions.

“The investment in Maxi Mobility is fully in line with our mission to promote concrete and scalable solutions for sustainable mobility. Maxi is tackling a crucial challenge with an innovative and pragmatic approach, making the transition to electric vehicles truly accessible and simple for taxi drivers and urban fleets. As Motor Valley Accelerator, we are excited to accompany them in this phase of growth and industrialisation,” commented Enrico Dente, director of Motor Valley Accelerator and Plug and Play.

With backing from both the finance and mobility sectors, Maxi Mobility looks to become a significant player in Europe’s sustainable urban transport landscape.

The post Maxi Mobility lands €1.2 million to electrify Italy’s taxis and accelerate zero-emission mobility appeared first on EU-Startups.

AI is transforming every possible industry, and VC is no exception. AI-powered tools streamline VC firms’ workflows, assist with research, and generate reports. They help investors process vast amounts of information more efficiently and make decisions faster. 

Indeed, it’s a powerful tool, but despite its advantages, it still falls short in some crucial aspects of venture investing. Assessing risks, understanding human behavior, and making strategic bets on the future all of these require human intuition, expertise, or at the very least, careful human oversight. Here’s why.

AI struggles with nuanced critical thinking and scepticism

AI models are great at aggregating information. Recent advancements in LLMs have significantly improved their reasoning capabilities by incorporating explicit chain-of-thought processes and citation tracing. However, despite these advancements, human judgment and oversight remain essential. AI often lacks the natural skepticism necessary for effective decision-making in VC, where subtle nuances matter. You have to deliberately instruct them to approach data with a pinch of salt; otherwise, they tend to smooth over the edges. 

For example, using AI to summarise pitch decks makes them look structured and polished. Ironically, that’s not always a good thing. AI-generated summaries strip away critical signals. They make everything look investor-ready. What’s more, AI tends to be overly optimistic about any pitch deck it analyses. 

VCs need to see decks in their raw, unprocessed form. A messy or poorly structured deck can speak volumes about a founder: how they think, what they prioritise, and how they communicate. Some VC-focused tools now incorporate “context-preserving” summarisers and link the summary back to the original deck. This feature helps preserve important nuances, but even then, careful human review is still essential. 

That’s why, for now at least, AI isn’t a go-to tool for analysing pitch decks, and probably it shouldn’t be.

If AI doesn’t know something, it still tends to make things up

Many models won’t admit when they don’t know something. They’ll just fabricate things instead and offer false or misleading data. For venture investors (and for other professionals as well, I guess), this is a serious problem. Due diligence requires verifying claims, and AI can’t be trusted to do this independently.

Some models, like Perplexity, try to mitigate this, openly stating when it lacks information and supporting its claims with sources. OpenAI’s Deep Research is also making efforts to produce reliable insights backed by references.

However, even with these models, there’s always a risk of making decisions based on misleading information. That’s why human oversight is still important.

AI has limitations in handling real-world data, messy and fragmented 

VCs deal with an overwhelming amount of information: from reports and financial models to emails and pitch decks. The data we operate on is fragmented and formatted differently: Excel files, PDFs, PowerPoint slides, Slack chats, notes, you name it. 

AI can now handle fragmented, multiformatted data much better than it could a few months ago. However, deeply understanding context and interpreting subtle signals and connections still relies heavily on human judgment.

But here’s the good news: AI is just getting started. Despite all these limitations, I believe the future of AI in VC workflows is promising. 

The recent viral video of two AI chatbots chatting in their own language, Gibberlink, offers a glimpse into what’s possible. One day, AI investment agents might be able to negotiate deals independently. 

While AI is not yet capable of making investment decisions, advancements in AI-driven automation continue to accelerate. One of our portfolio companies is developing AI-driven agents to support VC workflows. This could potentially add AI members to investment teams. Another founder I know is working on an AI agent designed to streamline communication with portfolio startups and partners. This agent integrates various business tools to automate relationship management, data collection, and company monitoring. Crunchbase recently announced another promising initiative: experimenting with AI to predict startups’ success. They’ve introduced an AI-powered ranking system to identify potential IPO candidates. 

However, I think the real transformation won’t come from AI replacing investors. It will come from AI enhancing them. Firms that adopt AI wisely, integrating it as a tool to automate workflows, will ultimately succeed.

The post Why AI can’t replace humans in Venture Capital – At least, for now! appeared first on EU-Startups.

Google says it will sign the EU’s AI code of practice, marking a fillip for the bloc, but the US search giant expressed concerns that the code risks slowing down European AI progress.

The voluntary code of practice lays out guidelines as to how AI model companies should comply with the EU’s AI Act, which comes into force next month, including legal responsibilities and copyright protections for creators.

It is not legally binding, but the EU says AI model providers who sign it will get increased legal certainty and reduced administrative obligations.

In a blog post, Kent Walker, president of global affairs and chief legal officer at Alphabet, Google's parent company, said he hoped the code “will promote European citizens’ and businesses’ access to secure, first-rate AI tools as they become available. Prompt and widespread deployment is important”.

However, he said Google, which is behind the Gemini AI chatbot, was "concerned that the AI Act and Code risk slowing Europe’s development and deployment of AI".

He added: "In particular, departures from EU copyright law, steps that slow approvals, or requirements that expose trade secrets could chill European model development and deployment, harming Europe’s competitiveness."

Google joins US AI firms OpenAI, Anthropic and French AI firm Mistral in signing the code, but Meta has said it won’t sign the code, warning that “Europe is heading down the wrong path on AI”.

Meta said: “The code introduces a number of legal uncertainties for model developers, as well as measures which go far beyond the scope of the AI Act.”

Microsoft told Reuters that it would likely sign the code.

UK-based AI startup Meet-Ting, also known as Ting, has raised €289k in an oversubscribed SEIS round backed by a group of prominent British entrepreneurs to develop its AI-powered scheduling assistant designed to book meetings directly through email.

“Imagine a world where meetings book themselves, people meet faster, and you never have to deal with endless emails back and forth or those link-based scheduling tools that make your guests do all the work. That’s the future we’re building with Meet-Ting,” says Dan Bulteel, CEO and Founder of Meet-Ting. “It lives in your inbox – just CC it – and helps you seem thoughtful. Because how you meet says a lot about who you are. And when we meet, amazing things can happen.”

Founded in 2025 by Dan Bulteel, formerly Global Head of Social at TikTok/ByteDance and adidas, Meet-Ting was co-founded with the venture studio Cocreatd, led by Oliver Yonchev and James Lawson Baker. Both Co-founders bring experience from ventures like Flight Story and Social Chain AG, along with a portfolio spanning companies including Amazon, Apple, TikTok, Disney, and Google.

Ting was conceived out of frustration. After years navigating the corporate calendar shuffle, Bulteel experienced yet another flurry of rescheduled meetings. A simple message to Yonchev – “Surely AI can fix this?” – kicked off a sprint that saw Ting built and funded in just two months.

Since then, the team has attracted advisors from Netflix, Disney, Google, and TikTok to help shape the product’s roadmap.

The result is a tool that reportedly operates with minimal setup – under one minute – and currently integrates with Google Calendar, with Outlook support on the way. Unlike traditional scheduling tools, there are no dashboards, no extra logins, and nothing to install.

The assistant reads the email conversation, understands the intent, checks availability, proposes suitable times, and sends calendar invites. The goal is to reduce the administrative drag that often delays or derails meetings – particularly for Founders, freelancers, consultants, and small business owners.

“Thanks to AI, scheduling doesn’t have to be so rigid any longer,” adds Bulteel. “Software can now adapt to us – our routines, our changes, and our painfully dynamic calendars.”

Meet-Ting is currently rolling out its closed beta, with early access available via Product Hunt and a waitlist now open at meet-ting.com. Ting’s model rejects rigid, link-based scheduling tools in favour of a flexible, conversational approach that works within email threads. By CC’ing ting@meet-ting.com, users can have meetings proposed, booked, rescheduled, and confirmed without leaving their inbox.

The startup is part of the Google AI Startup Program, receiving up to €303k in credits over two years for Google Cloud and Gemini infrastructure. Built using Gemini, Ting positions itself not just as a bot but as an assistant that adapts to the dynamic nature of human schedules – a problem the founders argue current scheduling tools fail to address.

Looking ahead, Ting plans to introduce features like persistent memory, allowing the assistant to learn user behaviour and scheduling patterns to make increasingly personalised suggestions. For now, the team is focused on refining the experience within Gmail, expanding compatibility, and welcoming users onto the beta.

The post “Surely AI can fix this”: London’s Meet-Ting just raised €289k to set you free from calendar jail with its AI scheduling assistant appeared first on EU-Startups.

Sometimes — not often — I meet someone building technology so profound it could reshape the future of the internet. 

SASHA is one of those rare companies,  developing a privacy-first tool that redefines how we govern, interact with, and protect content online. 

By giving individuals control over their images and embedding accountability into digital sharing, it has the potential to fundamentally shift how we participate in the digital world.

The technology makes it possible to prove ownership of an image, and potentially identify the first-link leaker if an image is abused online — even if the image is manipulated or screenshotted. 

I spoke to Thomas Eriksson, CEO and founder of SASHA, to learn more.

From a personal privacy breach to public mission 

SASHA (short for Safe Share) was founded in 2020 and seeks to prevent and address online image-based abuse (IBA) and identity theft. SASHA’s business aim is to empower and support victim-survivors to take back control over their images and to hold perpetrators legally accountable for their actions.

The company was inspired by the experience of a close friend of Eriksson’s whose ex-boyfriend had leaked intimate images of her online. It was heartbreaking. 

Eriksson admits, “I wanted to help her. I said, “You should report it,” and started looking for solutions.

"I was shocked at how big the problem was — how widespread image-based abuse is — and how hard it was to get anything taken down. We went to the police, but they told us it was difficult to do anything because perpetrators hide behind fake accounts.

Platforms said they’d take it down if it were public, but in private messages? That was harder — too many barriers.

Eventually, we got the content removed, but within 10 minutes, it reappeared under a new fake account with slight alterations. And it happened again and again.“

“No consequence,  no control”: the problem SASHA was built to solve

Eriksson is a Danish concept developer and serial entrepreneur with a background in medtech, edtech, and digital innovation. He has created products and campaigns for both creative and commercial organisations, including Roskilde Festival — where he led the "Roskilde Fever" campaign during the COVID-19 lockdown — Odense University Hospital, and the Danish Centre for Learning Materials.

He wondered, Why hasn’t this problem been solved?

“Reverse image search works, but only on publicly available content. There are watermarking tools, but they’re either visible, fragile, or centralised. And none of them help with the burden of proof.

Eventually, I realised the issue boiled down to a lack of consequence. There are no consequences for sharing images without consent online. That’s the root.”

In the physical world, we’ve built systems — laws, social norms — that enforce consequences. Online? Not so much. 

“So I asked, 'What kind of technology could restore consequences?'”

According to Eriksson, first, we need to document the intent of a share.

“If I send a photo of my kids to my mother-in-law via Messenger, and it ends up elsewhere, I should be able to trace it back. Just like if I lend you my car and you give it to someone else who crashes it — you’re still responsible.”

Second, the tech had to be robust. Images online go through compression, rotation, filters, and memes. 

“Metadata gets stripped the moment you upload. Most watermarking dies if you rotate the image 11 degrees or draw on it.”

Third, it had to be scalable. 

“We share 14 billion images a day. So we have to build something that works on the edge — on the user’s device — not in the cloud. That protects privacy and scales infinitely.”

How SASHA works

With SASHA, the watermark — or signature — is embedded directly into the image pixels. 

Eriksson explained:

“We’ve hidden what we call “needles in a haystack,” where the haystack is the size of a football field, and the needles are practically invisible.”

SASHA is resilient against both compression and manipulation, maintaining data integrity. Its system ensures robust protection of images directly on-device through a decentralised network. 

“Even after compression, rotation, and screenshotting, our signal stays intact. In our beta, we can even detect the signature from a photo taken of another phone screen.”

Further, as image abusers become more sophisticated, so does SASHA. Its iterative learning process studies and processes new attacks, applying these insights to prevent future occurrences.

 Importantly, SASHA complies fully with strict privacy standards such as GDPR, in line with EU laws. It solely uses an imprint of each image for recognition, ensuring images remain private.

The company does not have access to or store the original image. Users can share their catalogued images through the App. 

If another user tries to share an image through the SASHA App, SASHA uses AI-embedded technology to see whether this particular image file is a derivative of an image file explicitly marked as “not shareable” by another user. In this case, the SASHA App blocks the share.

This means that if someone finds their image somewhere it shouldn’t be, they can scan it and know exactly who shared it. They can send a takedown request — or a cease and desist letter — directly through the app. It gives people the power to act, not just wait for a platform to maybe care.

Embedding digital consent at the point of share

SASHA is currently concentrating on the B2C model, but its ultimate goal is the B2B2C model, which will make the B2C model redundant.

According to Eriksson, “right now, content moderation is reactive. Someone reports an image, a moderator reviews it, and decides if it breaks policy. “

“That’s slow, expensive, and subjective.

With SASHA, platforms can just ask the image: ‘Should you be here?’ And the image answers: ‘No, I was intended for this person, in this context.’ Boom — delete.

No need to see the content. No need to guess. Just enforce the owner’s intent.”

The biggest shift for SASHA will be when the tech is embedded into platforms. The company is in discussions with some of the biggest tech players to build this into messaging and upload functions.

The B2B2C model aims to not only make it easier for victim-survivors to identify a perpetrator of nonconsensual image sharing but also to prevent such non-consensual sharing in the first place. 

Imagine if, when someone tries to forward an image, their app says: “This photo was shared with you privately and cannot be reshared.” As Eriksson shared, “it’s like embedding consent into the image itself.

The company is also is close dialogue with insurance and telco partners that offer digital safety services. They’ll sponsor SASHA premium access for their customers. And the company will offer APIs to platforms that want to integrate our protection.

The next step for SASHA is deepfakes. Eriksson explained that because SASHA’s system knows the origin and intent of an image,it can detect when content has been altered or used maliciously. Deepfakes often start with real content. We can trace it.

Embedding oversight into the infrastructure of trust

The team is acutely aware that technologies designed to protect users can also be misused — or unintentionally expose already vulnerable populations to new risks.

SASHA took the novel step of engaging Tech Legality, a consultancy firm specialising in human rights and digital technologies, to carry out an independent human rights impact assessment (HRIA) of the SASHA product. 

The assessment was tasked with analysing the role SASHA’s product could play in preventing and responding to IBA and identity theft through a human rights lens; identifying potential adverse human rights impacts and make recommendations to prevent or mitigate those, including risks to SASHA’s users and others impacted by SASHA’s products, in particular vulnerable groups, and journalists and political activists; 

For example, there is a risk that governments could approach SASHA with requests for specific user data, either under national laws which may not always comply with international human rights law, or by informal requests where there is no legal basis, but still pressure is exerted.

Further, in an increasing number of jurisdictions (e.g. Ghana, Malaysia, Bangladesh, Pakistan), homosexuality is criminalised, and this means that where SASHA users who are LGBTQ+ share intimate images of themselves, they may risk leaving an evidence trail related to their production of materials that are deemed illegal under national laws. 

In these cases, the SASHA technology could be used to identify specific users for the purpose of a criminal investigation. 

Mindful of this risk, Eriksson explained,  “If a regime came to us and asked who took a picture of a protest — we couldn’t tell them. Only the owner can unlock that.”

Further, the SASHA product has been designed with the intention that adult users can use it as an avenue to safely share images, including sexual images, while protecting their images from unwanted onward sharing. 

However, children globally also share sexual images of themselves, commonly referred to as ‘sexting’. In many jurisdictions, including in some countries in Europe, ‘sexting’ is criminalised as it is deemed to be selfproduction of child sexual abuse materials. This means that if children use the SASHA app, it could produce evidence that may be used against the child in criminal proceedings. 

Children may be incentivised to use the App because it promises the ‘safe sharing’ of intimate images, which implies that their images will be kept private and that children will be safe from criminalisation.

In response, SASHA is made available only to users aged 16+, in line with European privacy laws. 

Eriksson admits, “That was a tough decision. Younger teens are vulnerable, but legally we can’t offer it to them without their guardians.”

"We’ve designed the system so that we cannot provide identifying information for minors under 16, even if asked. Our terms of use make that clear. We want to prevent harm, not contribute to it.”

Eriksson admits, “If we succeed at scale, we’ll need oversight. We’re designing our system so that independent third parties — like EU regulators — can verify how data is used and accessed. The database of image signatures must not be misused or exploited.”

A new wave of tools turning digital evidence into justice

SASHA joins a growing wave of technologies designed not just to document harm, but to actively support justice, dignity, and agency in the digital age.

For example, in terms of offender identification and prosecution, TraffickCam is a mobile app that helps combat child sex abuse and human trafficking by letting travellers upload photos of hotel rooms they stay in. These images help build a searchable database used by investigators to match photos from trafficking cases with specific hotel locations, aiding in the identification and prosecution of offenders.

UK-founded eyeWitness to Atrocities equips human rights defenders, journalists, and investigators with tools to securely document and verify evidence of war crimes and human rights abuses. Its Android app captures photos, video, and audio with embedded metadata (e.g. GPS, time, motion), ensuring authenticity.

 All files are encrypted and uploaded to a secure repository with a verifiable chain of custody, suitable for legal use. eyeWitness then compiles this material into structured dossiers to support international justice efforts, working with courts, civil society groups, and global institutions.

This is just the beginning for SASHA, a mission-driven company changing the lived reality for many of the internet. The company’s 

“Honestly, I’m really proud. Not just of the technology, but of the team. We’ve brought on some of the world’s best cryptographers and watermarking researchers, including people formerly at Google. They could work anywhere, but they chose SASHA because they believe in the mission,” shared Eriksson.

“We’ve created a tool that gives power back to individuals. That embeds responsibility into digital content. That lets us build an internet where privacy, consent, and accountability are not just ideals, but defaults.”

Lead image: Thomas Eriksson, CEO and founder of SASHA. Photo: uncredited. 

Paris-based Retab, founded by engineers frustrated by the “broken” state of document AI, today announces the launch of its platform and a €3 million pre-Seed funding round to support platform development and community growth, as the company scales its infrastructure to meet rising demand from vertical AI startups and internal innovation teams alike.

The round was backed by VentureFriends, Kima Ventures, and K5 Global, alongside Eric Schmidt (via StemAI), Olivier Pomel (CEO, Datadog), and Florian Douetteau (CEO, Dataiku).

“People keep building demos that look like magic, but break the moment you put them into production,” said Louis de Benoist, Co-founder and CEO of Retab. “We lived that pain ourselves. Wiring up fragile pipelines just to extract a few fields from a PDF. We built Retab because it’s the developer-first platform we always wished we had.”

Founded in 2023 by engineers from Cambridge and École Polytechnique – Louis de Benoist, Sacha Ichbiah, and Victor Plaisance – Retab is an AI agent that builds document extraction pipelines. Its all-in-one platform claims to turn messy PDFs, handwritten scans, and more into clean, structured data, without stitching together brittle third-party tools.

Louis and his Co-founders cut their teeth building internal automation tools for document-heavy workflows in logistics. Over time, they realised their true value wasn’t in the output, but the orchestration layer they’d built to make the models work. That tooling became the foundation of Retab.

The platform reportedly delivers guaranteed performance through a system of intelligent checks and balances: 

• Self-Optimising Schemas: An AI agent automatically tests and refines instructions based on a user’s documents, maximising accuracy before the system ever goes live.
• Intelligent Model Routing: The platform is model-agnostic. It automatically benchmarks and routes each task to the best-performing model for the job, whether the priority is cost, speed, or accuracy. This can potentially make processes up to 100x cheaper than other solutions.
• Guided Reasoning & k-LLM Consensus: Retab forces models to “think” step-by-step and uses a consensus mechanism among multiple models to quantify uncertainty, acting as a powerful safety net to ensure trustworthy results.

“Retab is the OS for reliably extracting structured data,” said de Benoist. “It wraps the best models in a layer of logic that actually makes them usable with error handling and structured outputs. That’s what devs need if they want to build production apps, not just prototypes.”

Use cases from their customers include:

• A major trucking company used Retab and found the smallest, fastest model configuration that could meet their 99% accuracy threshold, lowering operational costs.
• A financial services firm uses Retab to extract specific quantitative metrics and qualitative risk factors from 200-page quarterly reports – a task that previously took a team of analysts days to complete.
• Others are automating claims processing, medical records, identity verification, and onboarding with minimal setup.

According to Florian Douetteau, Co-founder and CEO of Dataiku and investor in Retab, “the AI-fication of the economy depends on the capability to convert operations based on millions of documents into verified, structured data that autonomous systems can utilise. On a large scale, this process hinges on quality control, cost efficiency, and rapid implementation. The team at Retab understands this thoroughly and is uniquely positioned to solve it for the thousands of AI first companies that are emerging.”

Looking ahead, Retab is expanding its platform to apply the same extraction methods to websites and is launching integrations with automation platforms like n8n, Zapier, and Dify.

Retab is also building toward its long-term vision: to serve as the intelligent middleware layer between the world’s unstructured data and the AI agents that need to understand them.

The post French document AI platform Retab raises €3 million for the “race to automate the world’s paperwork” appeared first on EU-Startups.

Klarna has been granted a licence in the UK, which will allow it to launch a debit card and offer savings accounts to its 11m UK customers as it looks to compete against UK challenger and established banks by offering some banking services.

Klarna’s UK subsidiary has been granted an Electronic Money Institution (EMI) licence by the UK regulator, the Financial Conduct Authority.

The authorisation will allow Klarna to offer some banking services in the UK, but stops short of allowing it to offer a full suite of banking services such as lending, which it would be able to offer under a banking licence.

Klarna said it had nothing to share, when asked if it planned to apply for a UK banking licence.

Klarna, which is most well-known for its BNPL offering, is looking to position itself as a “global digital bank”.

The EMI licence authorisation will mean that Klarna can launch its debit card in the UK, which it is piloting in the US, and UK customers will be able to hold and manage funds in their Klarna accounts.

In Europe, Klarna already has a banking licence which is overseen by Swedish regulators.

Abby Vickers, Head of Klarna Financial Services UK, said:” This authorisation marks Klarna’s next big step in the UK—moving beyond flexible payments into everyday financial management.

"While traditional banks are still playing catch-up, Klarna is giving consumers a smarter way to spend—and now, to manage their finances too. This is what modern money management looks like: manage, spend, and get rewarded—without the hassle."

In May this year, Klarna reported its revenues were up 13 per cent year-on-year to $701m in Q1, but losses more than doubled to $99m at the BNPL fintech.

The UK’s AI Security Institute (formerly AI Safety Institute) has unveiled The Alignment Project, a £15 million project dedicated to studying and mitigating the risks of advanced AI.

With advanced models now displaying PhD-level reasoning in some domains, experts caution that current controls may be insufficient to manage future risks.

AWS will provide critical infrastructure support through free cloud credits for alignment experiments. Other partners include Anthropic, Amazon Web Services (AWS), Canada’s AI Safety Institute and CIFAR, Schmidt Sciences, UKRI, ARIA, and civil society groups.

Geoffrey Irving, Chief Scientist at the AI Security Institute, emphasised:

“AI alignment is one of the most urgent and under‑resourced challenges of our time. … Misaligned, highly capable systems could act in ways beyond our ability to control, with profound global implications.”

Secretary of State for Science, Innovation and Technology Peter Kyle added:

“Advanced AI systems are already exceeding human performance in some areas, so it’s crucial we’re driving forward research to ensure this transformative technology is behaving in our interests.”

The Alignment Project operates through a three-pronged approach:

• Grant funding: up to £1 million per project across disciplines.
• Compute access: £5 million in AWS cloud credits for large-scale research.
• Venture capital: investment support for startups working on alignment solutions.

The launch comes as generative AI continues to reshape industries, from enterprise software to healthcare and education. At the Seoul AI Summit in May 2024, the UK and others agreed on a network of AI safety institutes -including Canada, Japan, and the EU- to share safety research and standards.

In parallel, the UK signed a strategic agreement with OpenAI in July 2025 to boost AI security and infrastructure collaboration, including support for deploying AI in public sectors alongside alignment research.

Governments, philanthropists, and venture firms are now being invited to join the project, either by funding research, offering compute resources, or investing in alignment startups.

Tzafon has closed a total of $9.7 million in a pre-seed round to significantly expand its computing infrastructure, accelerate the development of its autonomous AI agent technology, and launch Lightcone, an agent designed to operate seamlessly on a user’s behalf across any app or platform.

Tzafon is an AI company focused on developing action-oriented AI agents that can automate complex digital tasks. Its advanced multi-agent systems are designed to enhance operational efficiency for both enterprises and individual users.

At the core of Tzafon’s technology is a unique framework that enables agents to operate across digital environments. By integrating advanced machine learning, reinforcement learning, and a proprietary AI architecture, Tzafon is pushing the boundaries of what agentic AI can do. The company’s first model has already demonstrated industry-leading performance, achieving top results on web-based tasks in OSWorld.

While AI has made significant progress in areas like research, writing, and graphic generation, its capacity for meaningful action is still in its infancy. Tzafon is working to close that gap, bringing powerful, action-capable AI into real-world applications.

Noah Löfquist, CEO and co-founder of Tzafon, shared: 

Our vision at Tzafon is to build agents from the ground up that are capable of collaborating across any digital and, in the future, physical environment. This round enables us to significantly expand our compute resources to train smarter, better, and more agentic AI models. And launch Lightcone, which will reduce the friction between intention and action across any digital device.

The extension round was led by HV Capital, with participation from Streamlined VC, Kakao VC, Oliver Jung, and angel investors from OpenAI and xAI.

Lina Chong, Partner at HV Capital, commented: 

It's inevitable that AI will go beyond read / write and move into the realm of actions. There's a huge gap in the market with many infrastructure challenges to make agents lightning fast, scalable, and more autonomous when it comes to performing tasks. It's an emerging and fast-changing space, as is with all other aspects of AI. The Tzafon team continues to release agents that are highly performant and able to function cross-platform. An amazing feat for a team that’s just getting started. 

With the new funding, Tzafon aims to quickly scale its compute infrastructure to meet rising demand, speed up product development, and reinforce its strategic position. Soon, users will be able to access Lightcone directly and tap into cutting-edge agentic technology.

Our approach fundamentally reshapes how people leverage AI, providing not just incremental improvements but transformational results. 

Löfquist added.

Due, a borderless payment startup founded by Revolut alumni, has extended its Seed round to €6.3 million to launch their new API platform, enabling businesses to seamlessly access stablecoin payments as blockchain technology reshapes global finance.

The round was led by Speedinvest, as well as participation from Semantic, Fabric Ventures, Strobe Ventures, Polymorphic Capital, and other investors, to further Due’s mission to allow anyone, anywhere to access borderless payment rails. This more than doubles its original €2.8 million raise.

Robert Sargsian, Founder & CEO of Due, says: “Moving money internationally still feels like sending a fax in the age of streaming. With Due, a single API line makes the world one currency zone, delivering real-time settlement at fair FX that’s five to ten times cheaper than legacy wires.”

Founded in 2022, Due empowers businesses with a seamless borderless account: a multi-currency account to send and receive funds directly between local and foreign fiat currencies, as well as stablecoins.

Alongside this, Due offers an embeddable API designed for FinTech companies. Whether Founders are building a payments platform, payroll system, crypto wallet, or neobank, they can reportedly access, send, receive, and settle global fiat and stablecoin transactions with a single integration. The company says their API embeds cross-border financial infrastructure directly into a customer’s platform, allowing them to scale faster, unlock new markets, and deliver a seamless global experience to users.

In both cases, Due offers real-time FX and settlement in 80+ markets with a single integration. This is achieved by stitching together local payment rails, liquidity markets, and blockchain networks into one unified infrastructure.

All transactions are routed through stablecoin and on-chain rails where possible, enabling mid-market FX rates and offering savings up to 90% on routine transactions compared to traditional banking infrastructure. By removing the need for multiple banking relationships, currency accounts, or regulatory setups, Due aims to turn global treasury into a simple endpoint.

Today, over 500 companies use Due to move money globally, including:

• Rainforest Builder (an international reforestation company) uses Due to fund tree‑planting projects, sending USD and settling in GHS, SLE, ZAR and more.
• Several regional neobanks now tap Due’s rails to offer customers virtual EUR and USD wallets, unlocking seamless, low‑cost international remittances.
• Extended (founded by former Revolut team members and now one of the fastest‑growing perpetual exchanges) is building instant, compliant on‑ and off‑ramps for traders in more than 80 countries using Due’s platform.
• A Mexico-based importer pays Chinese suppliers same‑day through stablecoin rails, avoiding costly SWIFT legs.
• Neverless (a crypto investment platform with zero fees, also ex-Revolut) leverages Due to enable instant on-ramp in high-value, emerging markets, such as UAE and Mexico.
• Ledn (a leading digital lending platform) leverages Due’s payouts feature for loan distributions to its customers worldwide.

Olga Shikhantsova, Partner at Speedinvest, says: “Cross-border payments remain one of the most stubbornly broken pieces of global finance – and stablecoins are here to fix it. Due is solving this not with another app, but with infrastructure: a single, compliant API that simplifies transactions across both fiat and stablecoin rails. We backed the team early on and are excited to now lead this round as they scale the platform into a true global standard.”

The fresh capital will be used to extend and scale Due’s Global Stablecoin APIs, enabling faster, broader access to real-time settlement infrastructure. Due aims to increase its coverage to over 100 countries in terms of payments rails and currencies by the end of the year.

Due will also build more payment solutions, including invoices, cards and yield accounts, and hire engineering and compliance talent to support its exponential growth accordingly.

The post London-based FinTech startup Due secures €6.3 million and launches Stablecoin API for borderless payments solution appeared first on EU-Startups.

UK-based Vennre, a digital wealth creation platform providing access to private market investments, has announced the launch of Venture Capital Co-Investments. This first-of-its-kind offering opens the door to high-growth startup deals for qualified individual investors, with entry starting at just $5,000.

The launch represents a significant milestone in Vennre’s growth, following the platform’s achievement of over $25 million in Total Transaction Value (TTV) across both income- and growth-focused deals, underscoring a sharp rise in investor interest. With over half of investors choosing to reinvest and income distributions from all deals meeting or exceeding expectations, the platform’s strong rate of repeat participation reflects increasing confidence in Vennre’s curated, high-impact investment opportunities.

Designed for Vennre’s global community of HENRYs (High Earners, Not Rich Yet), including a growing segment of Shariah-conscious investors in Saudi Arabia and the wider GCC, this offering provides access to investment opportunities previously limited to institutions.

Ziad Mabsout, CEO of Vennre, commented:

Crossing $25 million in transactions is a proud milestone for our team - and a clear signal of growing investor confidence in our mission. With the launch of venture capital investments, we’re unlocking access to high-growth opportunities that were once reserved for institutions - especially for the next generation of qualified HENRYs in Saudi Arabia and across the globe.

The product is fully digital and Shariah-compliant, breaking down traditional barriers to venture capital investing, including high entry minimums, limited access, and a lack of transparency. It offers a streamlined, user-friendly experience tailored to the needs of next-generation investors.

This launch also reflects Vennre’s growing strategic focus on Saudi Arabia, aligning with Vision 2030’s goals of expanding individual participation in investment and wealth creation.

By providing access to curated, high-growth startup opportunities, Vennre is helping redefine what’s possible for qualified investors across the region, marking the first step in a larger mission to make private markets more intelligent, inclusive, and globally accessible.

Stockholm-based Tzafon today announced the final close of its €8.3 million pre-Seed funding round, to scale its computational capabilities, advance its autonomous AI agent technology, and launch “Lightcone,” an agent that can act on your behalf across any app or platform.

The extension was led by HV Capital, joining investors such as Streamlined VC, Kakao VC, Oliver Jung, and angels from OpenAI and xAI. The total round, according to the company, takes the top spot as Sweden’s largest pre-Seed financing.

“Our vision at Tzafon is to build agents from the ground up that are capable of collaborating across any digital, and in the future physical environment,” said Noah Löfquist, CEO and Co-founder of Tzafon. “This round enables us to significantly expand our compute resources to train smarter, better, and more agentic AI models. And launch Lightcone, which will reduce the friction between intention and action across any digital device.”

Founded in 2024, Tzafon is an AI company focused on building action oriented AI agents. Tzafon developsmulti-agent systems capable of automating complex digital tasks, enhancing operational efficiency for enterprises and consumers.

By leveraging cutting-edge machine learning techniques, reinforcement learning, and proprietary AI framework, Tzafon looks to take agentic AI to new heights. The company’s technology stands out due to its performance benchmarks with the first version of their model, reportedly achieving top results on web tasks on OSWorld.

The company outlines that while AI excels at researching, writing, and generating graphics, when it comes to taking actions, “we’re still at the starting line“.

“It’s inevitable that AI will go beyond read/ write and move into the realm of actions. There’s a huge gap in the market with many infrastructure challenges to make agents lightning fast, scalable, and more autonomous when it comes to performing tasks. It’s an emerging and fast changing space as is with all other aspects of AI.

“The Tzafon team continues to release agents that are highly performant and able to function cross platform. An amazing feat for a team that’s just getting started,” added Lina Chong, Partner at HV Capital.

With the new capital infusion, Tzafon plans to rapidly scale its compute infrastructure to meet growing market demand, accelerate product development cycles, and further strengthen its strategic market position.

Customers will soon be able to access Tzafon’s “Lightcone” product directly, enabling them to harness their agentic technology.

“Our approach fundamentally reshapes how people leverage AI, providing not just incremental improvements but transformational results,” Löfquist concluded.

The post Sweden’s Tzafon raises a total of €8.3 million to scale AI compute and take “agentic AI to new heights” appeared first on EU-Startups.

The race for AI dominance is taking an unexpected turn. While Silicon Valley burns billions chasing the next ChatGPT, European startups are quietly winning by going narrow.

In boardrooms from Stockholm to Berlin, a new playbook is emerging: forget building the next universal AI brain, build the smartest one for your industry. European AI startups pulled in 55% more investment year-on-year in Q1 2025, according to Dealroom, even as global tech funding contracted. Nearly half of Europe’s new unicorns were AI companies, but here’s the twist: they weren’t trying to out-GPT OpenAI.

Instead, they were laser-focused on solving specific problems like diagnosing rare diseases, automating legal due diligence, or optimising supply chains. The message is clear: in a world obsessed with generalist AI, the specialists are cleaning up.

If you’re building in health, legal, or mobility tech, your competitive edge won’t come from scale; it’ll come from depth. While the giants chase everything, Europe’s AI startups are proving that knowing one thing exceptionally well beats knowing everything adequately.

The great pivot: From universal to vertical

European AI strategy represents a fundamental shift from the scale-first mindset that dominates the U.S. market. Rather than pursuing ever-larger models, many European founders are prioritising precision over breadth – designing tools optimised for defined business contexts.

General-purpose models often struggle with integration, contextual relevance, and compliance constraints in enterprise-grade environments – areas where vertical AI excels.

According to MarketDataForecast, the European AI market is projected to grow from $119.8 billion in 2025 to $1.433 trillion by 2033. An industry analysis by Roland Berger noted that the most valuable AI companies are not building general-purpose systems but applying AI to specific domains with measurable value. According to Atomico, a significant share of AI funding in 2024 went to early-stage companies focused on vertical applications.

Regulation as a competitive advantage

Europe’s often-cited regulatory strictness is proving to be a catalyst for differentiated innovation. The EU AI Act emphasises ethics, explainability, and transparency, which has forced startups to bake compliance into the architecture of their solutions, offering trust and clarity as core product features.

One example is Paris-based Orasio, which raised €14 million in 2025 for a platform focused on AI-driven real-time video analysis in public safety. The startup’s solution integrates privacy safeguards and policy-aligned features, setting a precedent for AI systems that balance efficacy with accountability.

Another is Legora, a Stockholm-based company developing collaborative legal tech platforms built specifically for regulatory-heavy workflows in law firms. These tools are not attempting to serve everyone; they’re engineered for very specific professional use cases (EU Startups). 

Industry-specific innovation

European startups are excelling in areas where domain expertise matters more than compute scale. London-based Latent Labs, for example, recently raised $50 million to support AI-driven protein design, a high-impact application requiring a fusion of biology and machine learning expertise.

In manufacturing, vertical AI enhances existing systems rather than replacing them. Startups are embedding intelligent automation into production lines, logistics chains, and quality assurance protocols, delivering efficiency gains without disrupting established workflows.

Specialised systems can be designed to fit directly into established operational frameworks, minimising disruption and accelerating deployment cycles.

The capital efficiency: Smaller rounds, bigger impact

European startups are proving that narrow, high-precision solutions can gain traction and deliver business value with significantly less capital. In 2024, AI represented roughly 20% of all VC investment in Europe (€8 billion), with 70% flowing into seed through Series B rounds (TechCrunch).

France emerged as a regional leader, accounting for over €1.3 billion in AI investment,  driven in part by players like Mistral AI, whose offerings are shaped by multilingualism, regulation, and enterprise integration rather than raw model scale.

Europe’s comparative shortage of hyperscale infrastructure has become an unexpected asset. With only 18% of global data centre capacity (compared to 37% in the U.S.), startups are forced to prioritise efficiency over brute-force compute. Without hyperscale infrastructure, European startups have been forced to design more efficient inference architectures, a constraint that has driven innovation at the system level.

Talent density and domain clusters

Europe hosts approximately 120,000 AI professionals compared to 112,000 in the U.S., with talent distributed across research institutions and industry clusters, creating localised centres of excellence. Germany’s strength in automotive, France’s in aerospace, the Nordics’ in energy, and the UK in finance each provide unique ecosystems for vertical AI development.

The next chapter of AI adoption

As the global AI narrative shifts beyond model breakthroughs, Europe’s vertical AI approach is delivering real business outcomes. The newly announced €150 billion EU AI Champions Initiative explicitly backs specialised applications and enabling infrastructure over foundational model development, affirming the EU’s strategic direction.

Customers, too, are recalibrating expectations. The central questions are no longer “Can AI do this?” but “Can it do this reliably, compliantly, and cost-effectively?” For European startups, that is precisely the opportunity.

Specialisation as long-term defensibility

Specialised AI systems offer more than short-term product-market fit. They create durable competitive advantages by embedding regulatory compliance, domain-specific knowledge, and localised workflow barriers that are harder to replicate than model architecture alone.

The leading European AI teams are architecting systems that reflect regulatory realities, domain-specific complexity, and operational constraints, not just product ambition. This alignment may ultimately prove more defensible than any technological edge.

As the hype around general-purpose AI continues to dominate headlines, European AI startups are quietly shaping the infrastructure that will define the next era of applied intelligence. Their focus on specialisation, integration, and measurable value reflects both necessity and insight. And as global markets evolve, this measured, deeply contextual approach could position Europe not just as a participant in the AI race, but as a leader in its most impactful domains.

The post Europe’s quiet AI advantage: Why specialisation is outpacing scale appeared first on EU-Startups.

Ore Energy, a Netherlands-based energy startup pioneering iron-air long-duration energy storage, today announced that it has successfully connected its flagship iron-air battery system to the electric grid in the city of Delft — the first iron-air system to be grid-connected and fully operational anywhere in the world. 

The pilot system is also the first long-duration energy storage (LDES) solution to be entirely designed, built, and installed within the European Union. The first-of-its-kind deployment represents a significant technological milestone in long-duration energy storage and marks a defining moment in European energy sovereignty and resilience.

I spoke to Aytaç Yilmaz, co-founder and CEO, and Bas Kil, Business Development Manager at Ore Energy, to learn more. 

The core principle of iron-air long-duration energy storage is based on rusting and derusting iron. Yilmaz explained that “when the battery discharges, iron oxidises and forms a special type of rust.

“The system charges by using electricity to convert iron oxides (such as rust) back into metallic iron. During discharge, the metallic iron reacts with oxygen from the air to form iron oxides again, releasing electrical energy in the process.” 

“The core chemistry was first developed in the 1960s for use in electric vehicles. But with limited market demand at the time, the technology was shelved.

Today, with the rise of renewables and the urgent need for affordable, long-duration energy storage, it’s finally the right moment for this kind of solution.”

Why an iron-air battery?

Where lithium-ion peaks at 4 to 8 hours of storage, Ore Energy’s iron-air battery holds power for 100 hours or more, enabling multi-day load shifting, better integration of renewables, and reduced need for fossil backup. Unlike conventional batteries, which rely on scarce or flammable materials, Ore’s iron-air chemistry uses safe, abundant elements with no reliance on lithium, cobalt, or rare earths.

Ore Energy’s iron-air battery directly addresses several systemic challenges in Europe’s energy transition:

Reduces renewable energy curtailment. 

Europe is already wasting vast amounts of clean energy simply because there’s nowhere to store it when demand drops. 

Ore Energy’s 100-hour battery captures surplus power across multi-day periods, cutting curtailment by up to 44 per cent in modelled systems and helping shave billions off Europe’s energy bill.    -

Improves grid stability and reduces reliance on fossil fuel backup. 

Even modern grids with renewables like wind or solar still lean on gas-fired power during multi-day lulls in generation. These fossil fuel “peaker plants” are costly, carbon-intensive, and erode the economics of decarbonisation. Ore Energy enables renewables to meet demand without fossil backup, something current batteries can’t do.

Saves on overbuilding and grid upgrades. 

Without long-duration storage,  grid operators must overbuild renewables to ensure reliability, which inflates system costs and strains grid infrastructure.  Yilmaz shared that the company’s system is around seven to ten times cheaper than lithium-ion batteries, “primarily because we use iron and air — materials that are abundant, safe, and inexpensive. Among long-duration options, this is currently the most cost-effective chemistry available.”

By reducing peak capacity requirements and enabling time-shifting, Ore’s batteries cut system-wide costs by up to 63 per cent in modelled future energy systems. 

Eliminates foreign supply chains for full European energy independence. 

Utilities are under growing economic and regulatory pressure to reduce dependence on foreign-controlled supply chains, especially for critical materials and rare earths, which are heavily concentrated in China and conflict-prone regions. 

Ore Energy’s iron-air batteries are made in-house from abundant, EU-sourced materials. They are being built with a fully European supply chain, aligned with the EU’s Critical Raw Materials Act and energy sovereignty agenda.

Ore Energy’s 100-Hour battery tackles long-duration storage challenges

According to Kil,  the company is focused on three key application areas:

Co-location with renewables, particularly wind. 

“Wind generation is highly variable — you can get days of high or low output in a row. Lithium-ion’s short duration doesn’t help much in that context. But our 100-hour battery is ideal for shifting large volumes of energy over multiple days.”

Grid-scale flexibility. As utilities phase out gas and coal plants, they need new ways to match supply and demand. The batteries can help stabilise the grid as it transitions to renewables.

Data centres. Data centres consume enormous amounts of electricity and are under pressure to decarbonise.

“Our batteries offer them flexibility, cost savings, and the ability to make better use of renewable energy contracts.”

Ore Energy’s pilot system — which uses iron, air, and water to store clean energy for up to 100 hours — was deployed at The Green Village, a living lab for next-generation climate and energy technologies located at Delft University of Technology (TU Delft). 

The startup was spun out of TU Delft and maintains a close collaboration with the university. The nearby location allows for direct access and monitoring, while the site's microgrid setup provides ideal conditions for real-world testing and validation.

The installation is now collecting real-world operational data and will serve as a testbed for multi-day energy shifting, which is a key milestone on the way to full renewable grid integration. 

Ore Energy’s full-scale system will use modular 40-foot containers, each delivering up to 4.2 MWh of multi-day energy storage, optimised for low-cost, low-footprint deployment.

“This achievement is proof that Europe can lead the world in energy innovation and energy resilience. We’ve shown that breakthrough solutions like iron-air can move from lab to grid in just two years and can be built entirely with a European supply chain,” said Yilmaz. 

“Our battery doesn’t just store clean energy, it solves three of the grid’s biggest problems: it slashes curtailment, replaces fossil backup, and reduces the need to overbuild wind and solar.

Long-duration storage like ours is what makes renewable power reliable, affordable, and sovereign. And now it’s ready.”

“The Green Village exists to bring bold ideas out of the lab and into the real world. Ore Energy’s iron-air battery is exactly that kind of breakthrough,” said Lidewij van Trigt, Energy Transition Project Manager at The Green Village. 

“Connecting the world’s first grid-ready iron-air system here in Delft shows what’s possible when research, regulation, and industry align. We’re proud to provide a proving ground for technologies that will shape the future of Europe’s energy system.”

Market maturity has finally arrived, but regulations need to catch up

According to Yilmaz, iron-air batteries require a multidisciplinary approach — chemistry, mechanical engineering, and grid integration. 

“Until recently, the market wasn’t mature enough to support that kind of effort. But now, with the rise of wind and solar, the need for long-duration, low-cost storage is undeniable — and the conditions are finally right.”

The company plans to scale to 50 GWh/year by 2030. The company has set an ambitious goal of 50 GWh/year by 2030. To get to this, Yilmaz shared, “First, we need to scale up production rapidly. That’s already in motion. 

!But we also need market mechanisms — tenders, incentives, recognition of long-duration storage in energy planning. If regulation catches up, and the market embraces this technology, 50 GW is realistic.”

According to Yilmaz, the technology is ready, but the regulatory environment for long-duration energy storage is still evolving. There’s no standardised way to value or procure LDES yet, and that slows down deployment. 

“Things are improving, but policy support is essential to reach the scale we’re aiming for.”

Due, a London-based borderless payment startup founded by Revolut alumni, has extended its seed round to $7.3 million, more than doubling its original $3.3 million raise. The additional capital supports the launch of Due’s new API platform, enabling businesses to seamlessly access stablecoin payments as blockchain technology reshapes global finance.

Due enables businesses to operate globally with ease through its seamless, borderless account, a simple, multi-currency solution for sending and receiving funds across local and foreign fiat currencies, as well as stablecoins.

In addition, Due offers a powerful, embeddable API tailored for fintech companies. Whether building a payments platform, payroll system, crypto wallet, or neobank, developers can access, send, receive, and settle both fiat and stablecoin transactions through a single integration. This API embeds cross-border financial infrastructure directly into the customer’s platform, helping them scale faster, reach new markets, and deliver a streamlined global experience.

Both the account and the API provide real-time FX and settlement across 80+ markets, powered by an integrated network of local payment rails, liquidity providers, and blockchain infrastructure. The result is instant, low-cost settlement of digital assets and fiat currencies, unified behind a single, developer-friendly platform.

Due channels transactions through stablecoins and on-chain networks whenever possible, enabling mid-market FX rates and reducing costs by up to 90 per cent compared to traditional banking systems. By eliminating the need for multiple bank accounts, currency setups, or complex regulatory arrangements, Due simplifies global treasury into a single, streamlined endpoint.

Robert Sargsian, Founder & CEO of Due, says: 

Moving money internationally still feels like sending a fax in the age of streaming. With Due, a single API line makes the world one currency zone, delivering real-time settlement at fair FX that’s five to ten times cheaper than legacy wires.

Stablecoins, digital currencies tied to stable assets like the US dollar or the Euro, are transforming the way payments work. They enable money to move across borders with the same speed and ease as information flows across the internet. This shift is powering a new wave of payments platforms like Due, which offer around-the-clock, global transfers that are faster, cheaper, and more transparent than traditional systems, marking a clear break from outdated, cumbersome financial infrastructure.

Led by Speedinvest, the funding round supports Due’s mission to make borderless payment infrastructure accessible to anyone, anywhere. The round also saw participation from Semantic, Fabric Ventures, Strobe Ventures, Polymorphic Capital, and other investors, bringing Due’s total seed funding to $7.3 million.

Olga Shikhantsova, Partner at Speedinvest, says:

Cross-border payments remain one of the most stubbornly broken pieces of global finance – and stablecoins are here to fix it. Due is solving this not with another app, but with infrastructure: a single, compliant API that simplifies transactions across both fiat and stablecoin rails. We backed the team early on and are excited to now lead this round as they scale the platform into a true global standard.

The new funding will help Due expand and scale its Global Stablecoin APIs, making real-time settlement infrastructure faster and more widely accessible. By the end of the year, the company plans to extend its coverage to over 100 countries in terms of payment rails and supported currencies. 

Additionally, Due will develop new payment solutions, including invoicing, cards, and yield accounts, and grow its engineering and compliance teams to support rapid growth.

Lead image: Due team | Photo: Uncredited

After a year of building product and traction in stealth, Stackgini announced the closing of its pre-seed funding round in April 2024. Since then, the company has seen strong demand from DAX 40 companies and established enterprise clients such as Endress+Hauser, Grünenthal, Badenova, and Louis Motorrad, all of whom are already actively using its solution.

IT departments manage large IT portfolios (often over 1,000 solutions), while business units continuously submit new requirements. Manually coordinating these requests across departments, systems, and fragmented data sources slows decision-making and quickly results in redundant licensing costs and growing IT complexity. 

Stackgini addresses this challenge with a data-driven platform that enables companies to assess, select, and optimise their IT landscape more efficiently.

Johannes Bock, CEO and co-founder of Stackgini, shared:

We founded Stackgini to solve a problem every organisation faces: fragmented, manual, slow IT decision-making.

Our platform learns from existing IT stacks, contextualises new IT demands, and proactively promotes the reuse of existing IT solutions — with the speed a modern IT requires.

Stackgini is on a mission to fundamentally modernise IT demand management. Since closing its pre-seed round in April 2024, Stackgini has been developing its SaaS platform in close collaboration with its customers. The product leverages proprietary AI to analyse internal portfolio data, IT requirements, and market data, delivering actionable IT solutions in real time. Stackgini acts as a co-worker across IT governance, enterprise architecture and IT procurement teams.

The product has already gained the trust of leading enterprises.

Steven Waegenaer, Head of IT Governance and Strategy at Grünenthal Group, adds: 

For us, Stackgini is a real game changer in demand management: we are able to work together with the business units at an early stage and in a data-driven manner and make well-founded decisions - thus preventing unnecessary costs and shadow IT.

The pre-seed round was led by High-Tech Gründerfonds (HTGF) and xdeck Ventures, with participation from experienced angels including Julius Göllner, Dr. Niklas Hellemann, Frank Piotraschke, and Lukas Gottschick.

We backed Stackgini because the team is taking a fresh, scalable approach to a universal enterprise pain point,

says Maurice Kügler, Senior Investment Manager at HTGF, adding: 

With rising IT complexity and talent shortages in IT architecture and procurement, fueling IT decision-making with AI is a huge opportunity.

Markus Gick, Managing Partner at xdeck Ventures, adds: 

Stackgini has the vision, the team, and now the traction to redefine how enterprise IT decisions are made. We’re proud to support them on this journey.

As enterprise adoption accelerates, Stackgini is now focused on scaling its engineering, customer success, and go-to-market teams. The company is also strengthening integrations with existing governance, procurement, and enterprise architecture tools, driving toward a future where IT decisions take days instead of months. 

Our goal is simple: to become the AI-powered backbone of enterprise IT decision-making.

says Johannes.

Lead image: Stackgini team | Photo: Uncredited

Retab, an AI agent for building document extraction pipelines, has raised $3.5 million in pre-seed funding alongside the launch of its platform.

Retab is a developer platform and SDK that transforms document processing for the era of large language models. Developers simply define the data schema they need, while Retab manages everything else, from dataset labelling and evaluation to automated prompt engineering and model selection.

The idea for Retab was born from the founders' early work building internal automation tools for document-heavy workflows in logistics. They quickly realised that their true breakthrough wasn’t in the end results, but in the orchestration layer they had created to make AI models work reliably and efficiently. That layer became the core of Retab.

Louis de Benoist, co-founder and CEO of Retab, shared: 

People keep building demos that look like magic, but break the moment you put them into production.

We lived that pain ourselves. Wiring up fragile pipelines just to extract a few fields from a PDF. We built Retab because it’s the developer-first platform we always wished we had.

Today, Retab’s all-in-one platform is used by dozens of companies to convert messy PDFs, handwritten scans, and other unstructured inputs into clean, structured data, without the need for brittle third-party tools. Users simply define the data they need, upload their files, and Retab handles everything else: from dataset labeling and extraction logic to evaluation and benchmarking. It intelligently routes tasks to the best-performing model and automatically switches as newer, more effective models emerge.

Importantly, Retab isn’t just another large language model. It acts as the intelligence layer that makes cutting-edge models from providers like OpenAI, Google, and Anthropic usable for real-world, high-stakes workflows. By managing the full lifecycle of document extraction with verifiable accuracy, Retab enables teams to replace manual processes with fast, accurate, and self-improving workflows across use cases like contracts, invoices, and compliance documents.

The platform delivers guaranteed performance through a system of intelligent checks and balances:

• Self-Optimising Schemas - An AI agent automatically tests and refines instructions based on a user’s documents, maximising accuracy before the system ever goes live.
• Intelligent Model Routing - The platform is model-agnostic. It automatically benchmarks and routes each task to the best-performing model for the job, whether the priority is cost, speed, or accuracy. This can make processes up to 100x cheaper than other solutions.
• Guided Reasoning & k-LLM Consensus - Retab forces models to "think" step-by-step and uses a consensus mechanism among multiple models to quantify uncertainty, acting as a powerful safety net to ensure trustworthy results. 

Retab is the OS for reliably extracting structured data. It wraps the best models in a layer of logic that actually makes them usable with error handling and structured outputs. That’s what devs need if they want to build production apps, not just prototypes,

said de Benoist. 

With a lean team of just ten employees and a rapidly growing developer community, Retab is positioning itself as a core layer in the AI infrastructure stack, a tool designed not just to showcase what’s possible but to empower others to build with it.

Customers in logistics, finance, and healthcare are already benefiting from Retab. One major trucking company leveraged the platform to identify the smallest and fastest model configuration that met its 99 per cent accuracy requirement, significantly cutting operational costs. A financial services firm now uses Retab to extract detailed quantitative metrics and qualitative risk insights from 200-page quarterly reports, a task that once took a team of analysts several days. Other users are streamlining processes like claims handling, medical record processing, identity verification, and onboarding, with minimal setup required.

The round was backed by leading early-stage funds including VentureFriends, Kima Ventures, and K5 Global, alongside Eric Schmidt (via StemAI), Olivier Pomel (CEO, Datadog), and Florian Douetteau (CEO, Dataiku). 

Florian Douetteau, co-founder and CEO of Dataiku and an investor in Retab, noted that the broader adoption of AI across the economy relies on the ability to transform document-heavy operations into reliable, structured data that autonomous systems can effectively use:

On a large scale, this process hinges on quality control, cost efficiency, and rapid implementation. The team at Retab understands this thoroughly and is uniquely positioned to solve it for the thousands of AI-first companies that are emerging.

Looking ahead, Retab is expanding its capabilities beyond documents to apply its reliable extraction methods to websites. It’s also rolling out integrations with automation platforms like n8n, Zapier, and Dify to streamline workflows even further.

At the heart of Retab’s long-term vision is its goal to become the intelligent middleware layer between the world’s unstructured data and the AI agents that need to interpret it. Whether processing a loan file, a contract, or a customs manifest, Retab turns unstructured content into usable, safe, and programmable data.

The newly raised capital will fuel continued platform development and community expansion, enabling the company to scale its infrastructure to meet growing demand from vertical AI startups and internal innovation teams.

Lead image: Retab team | Photo: Uncredited