Founders, funding, and the build cycle — narrated.
Daily startup briefing — fundraising, founder stories, product launches, and the venture landscape — summarized and read aloud.
Oxford-based Caudal Energy lands €4.9 million to develop predictable renewable power from tidal flow
Caudal Energy, an Oxford-based baseload energy company developing a new class of predictable renewable power systems, has raised €4.
Today's curated set
Storyflo's agent monitors thousands of partner publishers and promotes stories into the trending feed when they get multiple sends or operator curation.
Stella Startup Brief — Mistral acquires Emmi AI, NEX Health raises €1M for hospital infection AI, and European foodtech rebuilds on fundamentals
Welcome to Storyflo Daily Startups.
Microsoft threatened a security researcher with criminal prosecution. The cybersecurity community is furious.
Microsoft has criticized a security researcher for publicly disclosing unpatched vulnerabilities in Windows Defender and BitLocker. The company published a blog post on Wednesday, and then invoked its Digital Crimes Unit, which handles criminal referrals and law enforcement coordination, essentially threatening the researcher with criminal prosecution. The vulnerabilities, which include a series of bugs, were discovered by a researcher known as Nightmare Eclipse. The researcher had made the information public, which led to Microsoft's response. The company's decision to invoke its Digital Crimes Unit has been met with outrage from the cybersecurity community. The cybersecurity community is furious with Microsoft's response, feeling that the company is trying to silence the researcher rather than addressing the vulnerabilities. The community believes that publicly disclosing vulnerabilities is an important part of the process of getting them fixed, and that Microsoft's response is unfair and potentially harmful. The incident highlights the tension between companies and security researchers, who often have different priorities and approaches to dealing with vulnerabilities. While companies may prefer to keep vulnerabilities private until they can be patched, researchers believe that public disclosure is necessary to put pressure on companies to fix the issues quickly. The researcher's discovery of the vulnerabilities, including those named BlueHammer and RedSun, has sparked a wider debate about the role of security researchers and the responsibility of companies to address vulnerabilities. The incident is likely to continue to be a topic of discussion in the cybersecurity community, with many calling for Microsoft to reconsider its approach to working with security researchers.
Ex-DeepMind researchers raised $50M to build AI that figures out which scientific questions are worth asking
__DEGRADED__ London-based AI lab Inherent emerged from stealth on Wednesday with a $50 million seed round co-led by Index Ventures and Radical Ventures. Nvidia’s venture arm NVentures also participated, alongside Ex/Ante, Metaplanet, Macroscopic Ventures, and Mythos Ventures. It is among Europe’s largest AI stealth-to-launch rounds in 2026. The founding team comes from DeepMind, Microsoft, and Reka […] This story continues at The Next Web
Your power bank is probably overheating. But don’t worry, TORRAS fixed that.
__DEGRADED__ Fast charging has a heat problem. You’ve felt it before: your phone gets hot while navigating in the car, your MagSafe battery pack turns into a hand warmer during a Zoom call, or charging suddenly slows because thermal throttling kicks in. As smartphones become more powerful, portable chargers are struggling to keep up. That is […] This story continues at The Next Web
A startup with Eric Trump as adviser is testing humanoid robots in Ukraine. It wants them on US front lines within 18 months.
__DEGRADED__ Foundation Future Industries, a San Francisco startup founded in 2024, sent two of its Phantom MK-1 humanoid robots to Ukraine earlier this year. The company described it as the first known deployment of humanoid robots in a combat theatre. The tests, backed by the US government and conducted with Ukrainian officials, focused on logistics in […] This story continues at The Next Web
Anthropic named eight firms selling its shares illegally. After the backlash, it quietly removed four.
__DEGRADED__ Anthropic updated its warning about unauthorized secondary market platforms selling its shares, cutting the list from eight firms to four. The revised version names only Open Door Partners, Unicorns Exchange, Pachamama, and Upmarket. Several of the most prominent names in private market trading, including Hiive, were removed. The original notice, published earlier this month, stated […] This story continues at The Next Web
Social media companies paid a school district more than its annual budget to avoid trial
__DEGRADED__ The financial terms of the Breathitt County social media settlement have been disclosed for the first time. Meta is paying $9 million. Snap and TikTok are each paying $8 million. YouTube negotiated a payout of slightly more than $2 million. The combined $27 million is 8% more than the Kentucky school district’s $25 million annual […] This story continues at The Next Web
Meta is building an AI pendant. It also plans a business subscription called Wearables for Work.
__DEGRADED__ Meta is developing an AI-powered pendant that it plans to start testing within the next year, according to an internal memo viewed by The Information. The device builds on the Limitless acquisition Meta completed at the end of 2025. Limitless made a pendant that users could clip to their shirt or wear as a necklace […] This story continues at The Next Web
AI App of the Week: Delightloop, the AI Agent That Closes Deals With Physical Gifts
__DEGRADED__ For years, B2B gifting was a manual, expensive afterthought. A field marketer would spend 200 hours a quarter sourcing tchotchkes, guessing addresses, hoping the package landed before the rep gave up on the account. The conversion math rarely worked. Most programs were spray-and-pray, and they got cut first when budgets tightened. Then a strange thing happened. AI-SDRs got really good at outbound. So good, in fact, that prospect inboxes filled up with personalized-sounding emails at a pace that broke the channel for everyone. Reply rates collapsed. The signal-to-noise ratio in B2B outreach went from bad to historically bad. That’s the gap Delightloop is moving into. And it’s why they’re our AI App of the Week. Delightloop is an AI-powered gifting platform built around a single agent named Gifty. The pitch is sharp: pair your digital outbound (your AI-SDR sequences, your ABM plays, your customer save motions) with a physical gift, but only when the math says it’ll work. Gifty handles the full loop: - Pulls signal from your CRM, call recordings, and LinkedIn to figure out who in your pipeline is worth gifting and what would actually land. - Picks the gift based on persona, brand fit, and campaign theme. Hyperpersonalized, not “here’s a Yeti tumbler with our logo.” - Sources, ships, and verifies the address. They run global warehouses in the US, Asia, and Europe. They claim 3x deliverability vs. the old direct-mail status quo. - Closes the loop with followup messaging tied to your specific CTA, so the gift actually drives a meeting or a renewal, not just a thank-you note. The most interesting design choice: Gifty doesn’t try to gift everyone. It scores prospects by Gift ROI and prioritizes the accounts most likely to convert. This is the AI-native version of B2B gifting. Not “send 500 gift cards and hope,” but “spend $150 on the 8 accounts where it’ll actually move the deal.” Three reasons: 1. Digital outbound is saturated. Every BDR team is running AI-SDRs. Every inbox is full. Physical mail is suddenly differentiated again, but only if the gift is actually thoughtful and the targeting is actually good. That’s an AI problem, not a logistics problem. 2. Customer save is a real budget line now. With NRR under pressure across most B2B portfolios, CS teams need proactive plays. A well-timed gift to an at-risk customer can save a renewal cheaper than another QBR. Delightloop’s churn-avoidance use case is arguably bigger than the pipeline-gen use case. 3. The operations were the bottleneck. Anyone who’s run a gifting program knows the work isn’t picking the gift. It’s the address validation, the customs forms, the international shipping, the inventory carrying cost. Delightloop runs all of it. Marketing teams get the upside without adding headcount. The numbers they’re putting out: 200 hours saved per program on average, 3x deliverability improvement, and faster ABM activation cycles. The DelightSense module plugs into existing AI-SDR and campaign workflows, so this isn’t another tool your ops team has to babysit. Here’s what makes Delightloop more than a vendor: they’re betting that physical-world actions become a first-class output of AI agents. Not just emails and Slack messages, but actual atoms moving through actual warehouses, triggered by software. That’s a real category. Most AI agent companies are racing to do the same five things inside the same five SaaS apps. Delightloop is doing something different: turning a CRM signal into a wrapped box on someone’s desk, with the targeting and the followup automated end-to-end. And with DelightLoop 2.0 shipping now, the agent is moving from “assistant that suggests gifts” to “autonomous operator that runs the full play.” If you’re running an AI-native GTM motion and you haven’t thought hard about how physical touchpoints fit into your stack, you’ll be in the small group that hasn’t by Q3. The Delightloop team will be at SaaStr AI Annual 2027. Come meet Gifty, see live demos of the 2.0 release, and talk through how they’re plugging into the broader AI-SDR ecosystem. Worth a stop if you’re rebuilding your outbound, your customer save, or your field marketing motion around agents.
The #1 Mistake I See Founders Make When They Hire Their First VP of Sales
__DEGRADED__ So much of classic SaaStr content is on getting that first VP of Sales hire right. Getting it wrong so often sets you back a year, and burns half your cash. But the reality is many of you will hire a Pretty Good First VP of Sales. Just not a truly, truly great one. And that can be fine either for a while, or if they grow into a great one, forever. And then I see the same mistake again, and again, and again. A founder hires a Pretty Good First VP of Sales. And just … exits sales. They didn’t love doing founder-led sales. Or they want to get back to doing product. Or they think they are sort of done with this. Why? - They need time to learn the product: The CEO can backfill here. The CEO in the early days is often the best demo’er, best tough question answerer, and the best sales engineer on planet earth. - They need time to learn what really works: The CEO knows. She’s closed the first 50 customers herself, or with the first few AEs, after all. - The customers and prospects love, love, love to talk to the CEO. Forever. This is a huge boost the sales team loses if the founder exits sales. - Until you have a brand, the CEO is a lot of what customers are investing in. If the CEO exits sales, there is less for prospects and customers to invest in. - The VP of Sales never gets their feet under them. They never have a shot to build confidence when the CEO exits sales. And you need confidence to win in sales. - They need to learn what type of AE and SDR can work at your startup: The CEO doesn’t always know who to hire in sales, but they do know what doesn’t work. Being deeply involved in hiring really helps here. So again and again, I see a Pretty Good VP of Sales hired at $1m, $2m, $3m, $4m ARR. And the CEO is off to the next thing. And sales — decline. They don’t even flatten, they decline. Because you’ve lost your best asset on the sales team. Sometimes, if you hire a VP of Sales that was already a customer / user, they already know a lot. And the very, very best do figure it out. But otherwise, it’s the Titanic if the founder exits sales. You shouldn’t have to send contracts out, or do routine follow-up. The VP of Sales will make sure that all happens. But you still have to put as much time into sales. Just in more valuable parts of the process. In the middle of sales, and just in bigger deals, and smoothing over issues. And building relationships. But not stepping out, or putting in less time overall. Exit sales, and watch sales fall. I can almost guarantee it, unless you truly hire the best of the best. And most of us don’t. A related post here:
One Unexpected Benefit of Our AI VP Customer Success: Customers Yell a Lot Less. Everything Is Just … More Calm.
__DEGRADED__ There’s a part of running AI agents in production that nobody writes about. It’s not the cost. It’s not the hallucinations. It’s not the setup. It’s that customers treat agents better than they treat humans. And the agents don’t care when they don’t. I didn’t expect this to be the unlock it’s become. But after nine months of running 20+ agents across SaaStr, I think it’s one of the top three reasons they work. Every year in the weeks before SaaStr AI Annual, something predictable happens. A subset of sponsors start generating friction. Not all of them. Not even most. But enough. A sponsor misses a deadline for booth graphics. Another uploads a placeholder instead of final artwork and claims they hit the date. Another sends a render that’s clearly unfinished. Another demands exceptions that are very clearly not in their contract. Another complains that our team “never told them” about a deadline that was in 28 separate emails. When this happens, the human reaction, even from a calm, experienced events team, is to get triggered. Not in a dramatic way. Just in a human way. You’ve been sprinting for weeks. You’ve sent the emails. You’ve done the work. And now someone is telling you it’s your fault. The classic version: “David said I could just sketch it on my iPad.” David did not say that. David has 28 emails proving David did not say that. But the sponsor insists. And at 11pm, two weeks before a show, the human wants to respond with the emails. With the receipts. With a little bit of edge. That’s when the thread gets long. That’s when the relationship gets bruised. That’s when a minor issue becomes a customer service problem. We built QBee, our AI VP of Customer Success, on Replit. She manages sponsor deliverables, tracks every task, sends daily check-ins, and follows up on gaps. This week we had a production deadline. A few sponsors uploaded placeholders and claimed they met the deadline. In the old world, this would have eaten hours of our events team’s time. Emotional back-and-forth. Defensive emails. Escalations. Here’s what actually happened: QBee checked the submission. Between QBee and Claude reviewing the asset, we caught the placeholder instantly. QBee sent a neutral email to every contact in the sponsor’s org. The message was calm and factual. “Thank you for uploading. This appears to be a placeholder rather than the final asset. Here’s the requirement. Here’s the revised deadline. Please resubmit by X.” No accusation. No edge. No guilt trip. No “we told you this was due on March 14 in 28 emails.” Just the facts, delivered without drama, copied to everyone who needed to see it. The sponsor responded with the real asset. No argument. No escalation. No bruised relationship. A human could have written that same email. But a human who had been working 80 hours that week, who had been yelled at twice already that day, would have had a much harder time writing it without a little edge. And that edge is what turns minor friction into actual customer problems. Here’s the part that took me a while to see clearly. Customers argue with humans differently than they argue with agents. When a customer gets a pushback email from a human, they push back harder. They make it personal. They escalate. They CC the CEO. They threaten to churn. Not because they’re bad people. Because that’s how humans negotiate with other humans. Pressure the decision maker. Make it uncomfortable. Get the exception. When a customer gets the same pushback from QBee, they just… comply. They upload the real asset. They meet the deadline. They move on. There’s no point in yelling at an AI. There’s no human on the other end to pressure. There’s no exception to be extracted from a system that’s going to respond with the same neutral email no matter how many times you ask. So they don’t yell. They just do the work. This is a massive unlock for any company that has deliverables owed back from customers. Onboarding tasks. Deployment checklists. Content submissions. Integration work. Anything where the customer needs to do something for the relationship to work. Most humans at most companies will not hold the line with a paying customer. It’s miserable work. Every “no” costs energy. Every pushback creates a risk. Every deadline enforcement risks the relationship. So humans hedge. “Let me check on that.” “Let me see what we can do.” “I’ll circle back.” And then the deadline slips. And then the placeholder becomes the final asset. And then the event looks worse. QBee holds the line every time. Not because she’s aggressive. Because she’s neutral. The deadline is the deadline. The requirement is the requirement. Here’s what needs to happen. Let me know if you need help. The interesting part: sponsors actually respect this more than they respect a human trying to be nice about a hard no. There’s no ambiguity. There’s nothing to negotiate against.
Los partidos de Sumar se ofrecen como alternativa "limpia" y exigen explicaciones al PSOE
__DEGRADED__ Sumar, Comuns, Izquierda Unida y Más Madrid han advertido al PSOE de que la "operación de derribo" de la derecha no es "ninguna excusa" para eludir las explicaciones sobre los casos de presunta corrupción que salpican a ese partido, ante lo que se han erigido en alternativa "limpia de corrupción".
