In short: The Trump administration’s FY2027 budget proposes cutting $707 million from CISA, eliminating the agency’s election security programme entirely and shedding 860 positions, a dramatic escalation that would reduce the country’s primary civilian cybersecurity agency to a $2 billion operation after a year already defined by DOGE-driven layoffs and mass departures. The United States’ central […]

In short: Google Maps now uses Gemini to suggest captions when users share photos of places, launching on iOS in the U.S. and expanding globally to Android in the coming months, the latest step in a six-month campaign to weave AI into every layer of Maps. Sharing a photo on Google Maps has always required a […]

In short: Paladin EnviroTech has acquired ICT, Ireland’s first R2v3-certified ITAD provider, completing a $70 million, nine-month acquisition spree that now spans the U.S., Netherlands, and Ireland, positioning the company to handle the growing wave of hardware disposal from Dublin’s hyperscale data centre cluster. When the servers that power Europe’s cloud infrastructure reach the end of […]

In short: Joby Aviation and Air Space Intelligence have announced a partnership to integrate AI-driven airspace management into U.S. electric air taxi operations, using ASI’s Flyways AI platform to model high-density eVTOL traffic before commercial flights begin later this year. The electric air taxi race has long centred on the aircraft itself: wing count, battery range, […]

For years, buying editorial coverage looked less like modern marketing and more like digging through email archives. A brand would hire an agency. The agency built a media list. Then came dozens, sometimes hundreds,  of outreach emails to publishers. Threads multiplied. Negotiations dragged on. And only if everything aligned would a sponsored article or press […]

A London-based startup which helps mitigate risks and vulnerabilities of businesses deploying so-called AI agents or AI tools that can complete specific tasks has emerged from stealth with a $13m seed round.

The funding round in Trent AI was led by LocalGlobe and Cambridge Innovation Capital, with participation from leaders at OpenAI, Spotify, Databricks, and Amazon Web Services.

The startup, founded in 2015, is tapping into the current trend of AI agent deployment, claiming its product is the first multi-agent security solution designed to secure agents as they evolve. AI agents and autonomous workflows introduce new security risks that traditional security tools were not designed to address, it says.

Trent AI says its product secures AI agents with specialised AI security agents that continuously scan environments, judge risk, mitigate vulnerabilities, and evaluate overall security posture.

The startup was founded by Eno Thereska (CEO), Neil Lawrence (chief scientist), and Zhenwen Dai (CTO), who are former Amazon/AWS and Spotify engineering leaders hailing from academia and research.

Thereska said: “Organisations are deploying AI agents and autonomous workflows faster than their security can adapt, and most development teams using these agents and workflows have no security framework designed for their systems.

“This is not an easy problem to solve. Trent AI is tackling these difficult and important problems, while building the necessary security foundations and frameworks for agentic systems now and through the next decade.”

The startup pointed to Deloitte research showing nearly three in four (74 per cent) companies plan to deploy agentic AI within two years, but that only one in five (21 per cent) report having a mature model for governance of autonomous agents.

Its product is built for developers and security teams that want to develop and ship agents fast without compromising security, it says.

Companies with early access to its product include Canopy and Weblogic.

Trent AI said: “These partners have reported: immediate visibility into their security posture, a security audit report, fast response time identifying and presenting vulnerabilities, a clean and well laid out remediation scope and adaptive feedback.”

The seed funding will support continued development of Trent AI’s security agents, expansion of the engineering team and growth of the company’s design partner and customer bases, it said.

The Berkeley biotech is backing a Nature-published approach that recreates the embryonic environment where blood stem cells first form, rather than reprogramming aged cells chemically or genetically. Its lead programme targets bone marrow transplant in blood cancers and has received FDA Orphan Drug Designation. HexemBio has publicly launched with a $10.4 million seed round led […]

The Los Angeles defence startup flew a demonstrator the size of an F-16 in March. A third aircraft is now in development. CEO AJ Piplica says the only way to build hypersonic aircraft at this pace is to accept that hardware will fail, and plan for it. Hermeus, a Los Angeles-based defence aviation startup developing autonomous […]

The London-based startup, founded by former BBC and Uber executives, runs AI-orchestrated video conversations that can gather structured insight from thousands of employees simultaneously. A seven-minute conversation produces more than 1,000 words of data versus ten from a typical survey answer. Natter, a London-based enterprise insights startup, has raised a $23 million Series A led […]

The Munich-based startup automates complex project workflows using AI trained on construction-specific data, not general-purpose models. Its backers include Earlybird, Pi Labs, noa, and Zacua Ventures. Conxai, a Munich-based construction AI startup, has raised €5 million in new funding. The round adds to the €2.7 million pre-seed the company closed in January 2022 with Earlybird […]

In short: Australian AI data centre company Firmus has raised $505m at a $5.5bn valuation in what it says is its final pre-IPO round, and is now targeting a $2bn listing on the ASX in June or July, backed by a $10bn Blackstone-led debt facility secured in February and a plan to deploy 1.6 gigawatts of […]

The venture debt facility, backed by InvestEU and signed on 7 April, takes PLD Space’s total 2026 fundraising to €210 million following its €180 million Series C in March. MIURA 5 is on track for its inaugural test flight later this year from the Guiana Space Centre. The European Investment Bank has signed a €30 […]

The French-Belgian medtech uses a portable headset to capture up to 800 infrared images per eye per second, extracting oculomotor biomarkers that can indicate Parkinson’s, Alzheimer’s, and multiple sclerosis years before clinical symptoms appear. It received CE certification in January 2025 and is targeting FDA clearance in 2026. neuroClues, a French-Belgian medtech company developing an […]

In short: Anthropic is in negotiations to anchor a new joint venture with Blackstone, Hellman & Friedman, and Permira that would embed Claude across private equity portfolio companies, investing roughly $200m of its own capital into a vehicle that could raise up to $1bn from buyout firms, and taking Palantir’s forward-deployed engineer model as its template. […]

Handhold, an AI-agent platform that turns visitors into customers without sales reps, has raised €3 million Seed funding led by Entourage Capital, with participation from Inovia Capital and e2vc. Angel investors include:

• Markus Villig (founder of Bolt),
• Ott Kaukver (former CTO of Twilio),
• Harsh Sinha (CTO of Wise), and 
• Janer Gorohhov (founder of Veriff).

Software buyers are more informed than ever and expect instant, personalised support. However, in reality, they often face unnecessary demo calls, generic onboarding flows and multiple handovers. When selling software, companies choose between product-led and sales-led motions. The choice comes down to setup complexity, product depth and price point.

But when the factors don't align, neither approach works cleanly: product-led growth leads to low activation and high churn, while giving a dedicated sales rep to every SMB account isn't economically viable.

Handhold automates the inbound customer journey, from first interaction to renewal, using AI account managers that qualify leads, run demos, and activate users around the clock.

I spoke to Georg Vooglaid, co-founder and CEO of Handhold, to learn more.

Why high-touch account management doesn’t work for most customers

Originally from Estonia, Handhold co-founder and CEO Georg Vooglaid started his career at TransferWise, then moved to London to study and later joined Seedcamp. ​ But while advising founders, he realised he wanted operating experience of his own. He found it at identity verification company Passbase, where he worked across the revenue organisation as the business scaled across New York and Berlin. There, he built close relationships with the company’s largest clients and supported them from validation through retention and upsell. The model proved effective, but only for the top few accounts where the economics justified that level of personal attention. ​

He explained:

“With our biggest client, I was available on WhatsApp 24/7. If we could’ve, we would’ve liked to give that level of attention to every customer, but there was no way we could afford to staff it.” ​

After leaving Passbase, Vooglaid reunited with his now co-founder to build a compensation benchmarking side project, which they later sold. ​

In May 2023, they launched what would become Handhold, initially as a conversation intelligence tool used by companies including Superhuman, Moss, and Katana. But after running into the limits of long sales cycles and relatively low pricing, they pivoted in late 2024. ​

The result was Handhold’s current product, shaped by customer feedback and his earlier experience. Turning every prospect into a managed account

According to Vooglaid:

“With AI agents, we can now replicate that one-to-one experience at scale because the economics work. You can give a ten-person startup a personal account manager that works around the clock, in any language, without actually growing headcount.”

Each prospect or customer is assigned an AI account manager who validates use cases, runs demos, supports onboarding, and manages the relationship over time.

From funnel to full lifecycle

Handhold aims to manage the full customer lifecycle — from first website visit through activation, retention, and expansion. With Handhold’s agents taking care of advancing the top of the funnel and closing smaller clients, sales teams can fully focus on building relationships with mid-market and enterprise accounts.

The buyer feels like they’re dealing with the same account manager, but behind the scenes, three agents actually split the work: ​

• One handles text Q&A and lead qualification,
• Another runs voice demos tailored to the prospect,
• A third sits inside the product to tailor activation paths for new users.
  

While you don’t have to use all three agents, context carries between them, and each following agent adapts to what the previous one has already discovered. Vooglaid explained:

“We approach it from the customer journey. What does the user actually need at each step? If you have a quick question, text works best. If you want a deeper understanding, a demo agent works better.

During onboarding, the agent might sit inside the product and guide you directly in the UI. From the user’s perspective, it’s always the same agent — in our case, “Holly” — but the form adapts to the context. The agent is there when needed and invisible when not.”

Vooglaid contends, “If we can price based on the number of customers we manage end-to-end, that’s a strong signal that we’re delivering real value. It means we’re not just improving one part of the funnel, but actually driving outcomes across the entire journey.” This experience shaped his view on a broader problem:

What human–AI interaction really looks like in practice

I was interested in learning how people interact with AI agents. Vooglaid notes that people are initially very direct with agents — almost testing them.

“Only once the agent proves itself does the interaction become more conversational.”

Second, attention is critical.

“You have five to ten seconds to engage someone — there are no social norms keeping them in the interaction. Third, active listening matters more than we expected. Giving users space to process and respond significantly improves the experience.”

And finally, personalisation is powerful, but risky. He contends that good personalisation can close a deal, but if you get it wrong and make incorrect assumptions, it can completely break trust.

Vooglaid explained that Handhold started at the top of the funnel — pre-sales, which enables the team to gather context early.

“For example, if a user interacts with a Q&A agent first, that context carries into the demo. The demo is then tailored to what they’ve already asked.

As we expand deeper into the journey, that context compounds. At the same time, users become familiar with the agent, which builds trust. So when the agent appears later inside the product, it doesn’t feel intrusive.”

More agents, but one winner: the account manager layer ​

Agentic sales agents and tools are likely to become more commonplace, even with large players like Salesforce, which acquired Qualified. Vooglaid sees the real competitive battleground in building a persistent AI account manager that supports customers across the entire journey.

“We’re particularly focused on SMB segments, where traditional sales models are harder to scale economically. That’s where this approach delivers the most value.”

The core benefit for customers is conversion and efficiency. Vooglaid explained:

“Some companies use us to handle inbound volume they can’t manage manually. Others use us for lower-priced segments where it doesn’t make sense to allocate a salesperson.”

Handhold also sees strong demand from companies operating across time zones — the agent can handle traffic outside working hours. Ultimately, the goal is to convert more visitors into customers without compromising the experience.

Where AI agents work today and gain the most traction

Handhold soft-launched in September 2025 and grew to a strong six-figure ARR run rate by year-end. The platform is working with more than 15 customers, who are seeing strong early results. For instance, workforce management company Parim reported a 60 per cent reduction in bad-fit demos alongside double-digit month-over-month growth in sales-qualified leads since deploying Handhold.

Vooglaid admits much of this is still trial and error. One advantage, however, is access to large volumes of sales call recordings and transcripts.

“From those, we can extract best practices — how to handle objections, how to structure demos — and use that to improve the agent’s performance.”

In terms of customer response, while people are already familiar with agentic Q&A and onboarding, the demo agent is more novel. Vooglaid admits that early on, there was concern around quality and whether the experience would feel polished enough.

“We’re now seeing more success with 'reverse demos,' where customers experience the agent themselves first. Once they see it working well, they’re much more open to deploying it.”

As for the hardest parts of the customer journey to get right, Vooglaid points to qualification and routing, detailing: “If companies position the agent as a core part of their brand experience, the quality bar is much higher."

"They need it to perform really well. If the alternative is to enter new markets or cover gaps in availability, companies are more flexible.

So a lot of this comes down to brand risk and how central the agent is to the customer experience.”

From human sellers to hybrid sales models

As for job displacement, Vooglaid asserts that Handhold is not directly replacing roles today. Instead, he sees that companies want their sales teams focused on mid-market and enterprise customers.

“We help handle the lower-value segment more efficiently. For example, one UK customer reduced low-quality demo bookings by 60 per cent. That didn’t reduce headcount — it allowed their team to focus on higher-value opportunities.”

Looking ahead, Vooglaid believes tools like his are fundamentally changing how software is bought and sold. AI enables more personalised, scalable interactions, which opens up new business models. It also shortens the sales cycle.

“We may even move toward a world where agents are buying software on behalf of users, or where some purchases become nearly instant. Exactly how this evolves is still unclear, but the direction is toward faster, more efficient buying.”

Handhold is already exploring use cases beyond SaaS, with clear applications across many industries. For example, the startup is piloting with a telecom company the use of agents to guide users through installing a Wi-Fi router. You scan a QR code, and the agent walks you through the setup step by step.

According to Pieterjan Bouten, Partner at Entourage Capital, AI agents are fundamentally changing how we buy software.

“From enabling new motions to becoming buyers themselves, Georg and Uku are building a system that opens up completely new business models and helps companies validate what’s working about their sales motion and what their customers need.

The early traction is a sign that they're on to something, and we’re excited to help them redefine how software gets sold in the future.”

The funding will accelerate go-to-market efforts and expand Handhold’s ability to move from converting leads to managing entire customer bases.

The AI design platform is paying creators based on engagement performance rather than audience size, marking its transition from a tool into a platform where creators can earn directly. The launch follows an AI agent marketplace the company introduced in March. Picsart has launched a creator monetisation programme open to all of its more than […]

UK-based analytics startup WholeSum has brought its total Pre-Seed funding to $1.3 million, with an additional $335,000 new investment from Love Ventures, Beamline, and strategic angels, following its initial $965k raise led by Twin Path Ventures announced earlier this year.

The round comes amid growing demand from enterprises in high-trust sectors, where organisations are increasingly finding that existing AI tools fail to deliver reliable, auditable insight from large volumes of text data.

While most organisational data is unstructured, teams continue to struggle to analyse it at scale. In practice, many have turned to LLMs, only to encounter hallucinations, inconsistencies, and outputs that cannot be reproduced or defended – particularly in regulated environments such as healthcare, financial services, and defence.

Founded by Emily Kucharski and Dr Adam Kucharski, WholeSum was born out of the founders’ frustration with existing AI tools while analysing large-scale qualitative datasets in a previous venture. The experience highlighted a systemic problem: organisations want to extract meaningful insight from qualitative data, but lack tools that are both scalable and scientifically defensible.

“From talking to dozens of large organisations making high-stakes decisions, we’ve seen a clear pattern: teams are experimenting with AI for text analysis, but quickly hit a wall when outputs can’t be trusted or reproduced,” said Emily Kucharski, cofounder and CEO of WholeSum.

“This funding allows us to move faster in building infrastructure for robust analysis at scale.”

WholeSum addresses this gap with a hybrid AI and statistical inference platform that converts free-text data into uncertainty-aware, reproducible, and auditable insight.

Designed as an API-first infrastructure layer, it integrates directly into existing analytics workflows, enabling organisations to extract nuanced signals and underlying drivers with the same rigour as numerical data.

Since its initial raise, WholeSum has seen strong traction across enterprise organisations in high-trust sectors. Early work with universities, financial institutions and pharmaceutical companies has demonstrated that the most valuable early signals are often buried in unstructured text data rather than in lagged quantitative metrics.

“Generic LLMs can’t deliver the consistent, reliable signals that high-trust industries need from unstructured data,” said Bill Corfield, Principal at Love Ventures. 

“Emily and Adam are uniquely positioned to solve this, and we're delighted to be backing them as they scale across Pharmaceuticals, Financial Services and beyond.” The company is now rolling out pilots and enterprise integration for increasingly complex, large-scale datasets.

The additional funding will be used for R&D, expanding the company’s world-class scientific and engineering teams, and scale enterprise deployments in sectors where methodological rigour is critical.

“All of us live in the dark most of the year, and we don’t have anything better to do than build,” says an organiser of one of a new wave of hacker houses which have cropped up in recent months across Scandinavia and the Baltics.

This resurgence in hacker houses can be seen all across Europe. But across Scandinavia and the Baltics, the hacker house mentality, the esprit de corps, is particularly acute, given the relative smallness of the markets.

Amid waning deal flow, particularly across the Baltics, investors see them as a quick-fire win to turbo-charge investment, while builders view them as a chance to showcase their talents.

Latvia-based Shipyard, Estonia’s ruum, Lithuania’s Basedspace and Denmark-based Bifrost House are examples of the new breed to emerge.

Hacker houses became popular in the early 2000s, particularly in Silicon Valley and the San Francisco Bay Area startup ecosystem- in response to rising housing costs and the need for collaborative startup environments. They are now enjoying a resurgence, with a more explicit hacker house/ builder lab format.

ruum

Tallinn-based ruum is one of the new crop of hacker houses which launched last year.

Helery Pops, ruum founder who is also a VC and angel investor, says ruum took inspiration from Basedspace, a neighbouring hacker space in Lithuania.

Pops said: “When we started looking into it, it was like an avenue of green flags everywhere. There was really no reason not to do it.”

The idea was to give free working space to builders just starting out in their careers, she says. Central to ruum's hacker house programme was a full-day hackathon, with 12 teams selected from 110 applicants, with the resulting progamme running for two and a half months.

Pops said: “There have been many hackathons, there have been many accelerators, there are some that are still on-going in Estonia. But at this point, it seemed that the deal flow was not coming on as it had been maybe in the last five years. There was space for something new. People are seeing there is a problem with the early-stage companies. This just seems like a possibility to solve a little bit of it.”

The programme was supported by €20,000 in funding, which came from Skaala, the family office of Skype alumni and Wise cofounder Taavet Hinrikus and his Skype colleague Sten Tamkiv, Startup Estonia, and a few angel investors.  

The broader Estonian tech community helped out by offering their services for free. The winners, Bilt.me (a "Lovable" for mobile apps), a six-strong team with an average age of 21, were whisked off for a week in San Francisco.

Pops says: “They work six days a week. If you are in the working space with them, and you start going home at 5pm, they are generally shocked. It is very cool to see how much energy and power they are putting into it.”

Ruum might have started out as a hobby, but Pops said the future could see more programmes, with a slightly amended format.

Shipyard

Meanwhile, across the Baltics, in Latvia, another newish hacker house space is the AI-centred Shipyard.

According to one of Shipyard’s founders, Marija Rucevska, who is also GP at a VC investing in the Baltics, Shipyard was created recognising the impact of AI on startups.

Rucevska said: “This new movement is acknowledging that you can move a lot faster to market and understand if it’s worth building something longer term or maybe just something that you are building for yourself.”

Like ruum, Shipyard’s admission programme was a 48-hour hackathon, with the programme then taking on 20 teams over three months, which is then whittled down further to a group of eight teams looking to get pre-seed funding.

Part of the programme, designed to turn builders into founders who are building AI-native teams, demanded builders deliver weekly shipping cycles.

“If you don’t deliver, you are out,” she says.

She adds: “We want to cherish and nurture the building, energy and spirit and also witness some of those of new AI native founders.”

Shipyard also helps Rucevska tap into her VC pipeline efforts, she said. She says: “We have really technically brilliant people here, a lot of very AI-savvy teams applying different types of tools.”

On the broader emergence of hacker houses across the Baltics, she said: “That kind of comes from the fact that all of us live in the dark most of the year and we don’t have anything better to do than build.”

She says the Baltics are also “very ecosystem driven” given the relative smallness of the countries, which means that individual countries are always rooting for each other.

Bifrost House

In Scandinavia, there is Copenhagen-based Bifrost House, which bills itself as “Copenhagen’s most ambitious startup community and co-working space”.

It is a venture studio that builds startups from the ground up, with a hacker house mentality. Bifrost House raises capital, forms founding teams, and operates businesses across sectors, including defence technology, consumer goods, B2B SaaS, and financial data. 

It is currently raising a €30m fund and has hitherto built 25 startups, with a mission to build 100 companies a year.

Sophus Blom-Hanssen, who runs the operation, said there is nothing like Bifrost House in Europe, given the scale of its ambitions.

He says: “The reason we are doing it, and the reason a lot of these type of spaces will emerge is that the whole modus operandi of a startup, of getting a startup to market quickly, that timeframe is compressing super-quickly. Both to build a product and the timeframe you are relevant in the market is also compressing.”

On its fund, he said: "We look at what does the business case we are scoping around require from a funding perspective?

“And work together with the founder and entrepreneur to structure a cap table that works to actually reach the first or second milestone so that the business can raise more money."

IMAGE: Bifrost House

The Iran war is not making cloud suddenly expensive. It is exposing how cloud economics was always downstream from energy markets, and Europe is structurally exposed. The current conflict in the Middle East is no longer contained to the region. Its effects are beginning to ripple into economies still recovering from the 2022 energy crisis, […]