News

There’s a lot of noise in enterprise AI right now. Under mounting pressure to deliver faster, safer digital services, businesses are turning to the next evolution in automation: Agentic AI.

No, this isn’t bolting on a chatbot and calling it digital transformation. AI agents are built to understand your organization, operating within your domain constraints with real autonomy. These agents operate inside your business, using your data to automate decisions, adapt to real-world problems in milliseconds, and embed themselves directly into operational workflows.

They blend the general reasoning power of today’s large language models with domain- specific intelligence grounded in company data. That might be clinical records, compliance frameworks, or engineering logs - whatever your business runs on. The result? Systems that take action: surfacing insights, automating tasks, and adapting based on your company policies and workflows.

Demand for automation is growing, as are expectations around compliance, transparency, and data governance, especially in Europe. Agentic AI offers a response to both: scalable intelligence, designed to work inside complex regulatory frameworks.

That matters in sectors like healthcare, manufacturing, and financial services, where data security, explainability, and reliability aren't negotiable. These aren’t markets where “good enough” is acceptable. Customers simply can not tolerate hallucinated responses or unreliable systems where their data hits the public domain.

Agentic AI is safer. Not because it’s slower or more cautious, but because it’s built for the environment it’s deployed into.

Agentic systems rely on a layered approach, with different types of agents operating across an organization:

1. Human assistance agents support real-time decisions, generating summaries, highlighting next steps, and assisting in code review or sales workflows. They keep humans in control while removing friction from day-to-day tasks.
2. Transactional agents manage system-to-system workflows. They handle onboarding, verification, or inventory reconciliation autonomously when appropriate, with escalation when edge cases arise.
3. Autonomous agents identify and solve problems independently. In domains like DevOps, logistics, or diagnostics, they monitor environments, anticipate failures, and act proactively rather than reactively to resolve these types of issues. These agents work in tandem rather than in isolation. Together, they form an intelligent layer across enterprise systems - learning, adapting, and coordinating actions in ways that were previously siloed or manual. True digital transformation across the enterprise.

Key to all of this is the use of custom vector databases. Vector databases enable AI agents to fetch relevant, security-controlled context from sensitive data without actually exposing that data in its original form to the agent. This is a game-changer for regulated industries. Rather than relying on generic training data from the public internet, this draws directly from the institutional knowledge inside your firewalls.

That means better accuracy, stronger compliance, and fewer surprises. It also means outputs that reflect your standards, rather than what’s statistically likely.

Agentic systems are already transforming highly regulated sectors in Europe. In healthcare, they reduce administrative overheads, improve triage, and accelerate innovation while protecting patient privacy. In manufacturing, they’re powering predictive maintenance, supply chain optimization, and real-time field service. Within finance, these agents enhance fraud detection, refine compliance, and provide hyper- personalized services.

Agentic AI adoption is particularly strong in regions with tighter data controls - namely France, Germany, and the Nordics - because these systems respect the boundaries enterprises are required to operate within.

These systems increasingly rely on serverless inference, which allows businesses to scale their AI infrastructure without wedding themselves to their maximum theoretical usage. That’s critical in Europe, where innovation budgets are often tight, and sovereign infrastructure matters. Agentic AI is being built to meet those regulatory requirements from day one.

Yes, Europe’s regulatory environment slows things down. But that friction forces better thinking. It pushes enterprises to build with trust, accountability, and explainability. Creating market conditions where sustainable AI can thrive.

GDPR, the EU AI Act, NIS2 and other regulatory frameworks define the standards by which responsible AI can scale. As US start-ups chase MVPs and launch before the proper guardrails are in place, European enterprises may end up with AI that’s more compliant and generally more effective in the long term.

Agentic AI marks a turning point in how businesses interact with their data and workflows. It moves beyond static automation to deliver systems that act, learn, and improve within the constraints enterprises define.

This is not a plug-and-play future. It’s a future that demands thoughtful design, domain- specific strategy, and an unflinching focus on outcomes. The rewards will be sustainable and significant for the organizations that build smart and scale responsibly. The hype in off-the-shelf, plug-and-play solutions will fade. Agentic AI infrastructure is built for the latest ways of working. Enterprises that invest now and build with intent will lead in the next stage for what’s next.

We've featured the best AI writer.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

• An external IT partner to Scania lost its login credentials through an infostealer
• The hackers used the password to access Scania and steal files
• They asked the company for money, and later offered the archive for sale

Swedish automotive manufacturer Scania has confirmed suffering a cyberattack which saw it lose sensitive customer data.

Security researchers Hackmanac found a new thread on a dark web forum, in which a database allegedly stolen from ‘insurance.scania.com’ was being offered for sale to an exclusive buyer for an unknown sum of money.

“hi guys. we hacked new target and selling full attachment of 'insurance.scania.com'. Full attached files is 34,000 and first time hacked + just will 1 hand sell,” the ad, published in both English and Russian, reads. “few pic attached with remarks (for no one cant copy and scam people).”

Save up to 52% off Lifelock Identity Theft Protection!

Your personal info is in endless places. And any one of them could accidentally expose you to identity theft. That's why LifeLock monitors hundreds of millions of data points a second for identity theft. LifeLock. For the threats you can't control.

Preferred partner (What does this mean?)View Deal

After the thread was posted, Scania confirmed the authenticity of the claims, saying it was breached in late May 2025 as part of a supply chain attack which originated at an external IT partner.

"We can confirm there has been a security related incident in the application "insurance.scania.com", the application is provided by an external IT partner," a Scania spokesperson said.

"On the 28th and 29th of May, a perpetrator used credentials for a legitimate external user to gain access to a system used for insurance purposes; our current assumption is that the credentials used by the perpetrator were leaked by a password stealer malware."

"Using the compromised account, documents related to insurance claims were downloaded."

Although the company did not detail what information was found in the stolen files, it’s safe to assume that it is sensitive, possibly financial, or medical. The number of affected individuals is also unknown for now.

After stealing the archives, the threat actor tried to extort Scania for money, reaching out on multiple occasions and demanding a ransom. Since it ended up offering the database for sale on the dark web, we can assume that the company declined the generous offer.

Via BleepingComputer

• Infostealers on the rise: the latest concern for organizational defenses
• Take a look at our guide to the best authenticator app
• We've rounded up the best password managers

The Great Cassette Rewind continues, as more affordable tape-playing options capitalize on the love for collectible physical media, which has only grown in the wake of the vinyl revival.

Earlier in 2025, We Are Rewind announced a metal Walkman-like portable player that can play cassettes over Bluetooth or with the best wired headphones, and then followed it up with the We Are Rewind GB-001 boombox reinvention.

And yes, this goes all out at being a boombox just the way you remember it. It's got a front-opening cassette deck, it's got physical dials and buttons, it's got a folding handle, it's got twin VU meters on the front, it's got two-way speakers on each side, and it's slightly heavier than is really convenient.

There are two notable feature swaps from the boombox you had a few decades ago, though: the radio tuner is out and Bluetooth is in for wireless playback, and the battery is a built-in rechargeable option instead of needing a dozen cells each the size of a large hamster.

It's got a guitar/mic input if you want to play or sing along with the music, and you can use this input to record to the cassette deck. Sadly, you can't record to tape from the Bluetooth connection, and I wish it actually had two cassette decks so you could go really old school and record between them, but what's here is probably the right balance for using it in the real world.

Speaking of which, I did get to test it out, though in a slightly odd situation: I was at a big audio trade show, and we couldn't test it in the side room it was being shown off in, because there was a constant demo of much larger speakers.

So we just took it out into the public area, put it on the ground, and turned up the dial until I could enjoy my music and everyone around me was annoyed – the authentic public boombox experience!

Actually, I suspect that people were more curious than annoyed – they were at an audio trade show, after all, and this thing really stood out among all the more traditional hi-fi that dominated the show.

There is a deep, fundamental satisfaction that comes from turning up the volume and seeing physical VU meter needles start to jump higher and higher up their range; a reassuring supplement to what your ears are telling you, that rocking is about to happen.

Despite being at a hi-fi show, the GB-001's sound is not exactly audiophile, which will not be a shock to anyone – and is not what we really want from it anyway. Because I was listening to it on a noisy trade-show floor, obviously my assessment of its sound is very much limited, but it is at least representative of using the thing in the real world, where other sounds around you will conflict with it.

It's heavy through the low-end, which is common among outdoor speakers, because bass is easily lost when traveling through open air. It seems to know what it's doing with this bass, which felt fairly controlled and lively – the risk with going strong on bass is that it starts to become flat and lands with a thud, but this felt like it had bounce.

The mids felt a little lost among the sound around me even when I'd turned the volume pretty high, but the treble comes across, and again I feel like this is what you expect from this kind of speaker in this environment – all the movement and conversation around me is heavily in the mid-range and is most likely to overwhelm the audio, but I could still hear all the core essence of songs, and definitely the beat.

I couldn't tell you much about its delivery of detail in that environment, but I genuinely do feel like this asking the wrong question of a boombox.

One thing to note is that it delivers a notably broader sound over Bluetooth than it does from cassettes. It's one big reason why I don't think the tape revolution will ever take off in the same way as the vinyl revival. I know some people love them, but really cassettes were the best solution to a portability problem at the time, and basically every solution since then has been superior.

But there is a genuine charm to them outside of the sound, which is the same with vinyl. The physical ritual of opening a box, sliding out the cassette, and sliding it into the boombox's caddy is deliberate and satisfying.

The chunk of pressing the mechanical buttons tickles the right part of the brain. The click and silence of a successful rewind delivers anticipation that you're about to dig into something good.

And I've written about this before, but I genuinely miss the creativity of the mixtape era, and both of We Are Rewind's products have that in mind, both including a line-in option to record something personal to tape and share it with your crush/best fried, or sibling, or favorite artist you waited ages at the stage door to see who's definitely going to love your stuff and will be your big break.

Even with all that acknowledged, it's the light-up twin VU meters that are my favorite part of the whole thing. They're another element that adds anticipation; when the lights come on and the needle twitches, you're about to kick off a good time.

My next home hi-fi upgrade will be genuinely influenced by the presence of features like this, because now that it's possible to play all the music in the world in high quality with almost no equipment whatsoever (just some of the best earbuds and your phone), if you're going to invest in big physical units to provide pleasing playback, they should make you as happy physically as they do sonically.

I think the GB-001 gets this – it's about the ritual of music, at home or out at the skate park.

• Bluetooth speaker heaven: my top 4 launches of 2025 so far, and what’s coming
• I listened to a $3 million stereo system, and I don’t know if I could hear every penny, but I definitely heard the dollars
• Everyone needs 5 pairs of headphones: my job is testing 'em – here are the ones to buy

• CISA flags security issue affecting multiple TP-Link models
• It allows threat actors to execute arbitrary system-level commands
• Affected models have all reached end of life, so should be replaced anyway

Multiple TP-Link routers, which have long reached end-of-life (EoL) status, are being abused in real-life attacks, the US government is warning.

The US Cybersecurity and Infrastructure Security Agency (CISA) has added a command injection vulnerability to its Known Exploited Vulnerabilities (KEV) catalog, signaling abuse in the wild.

A command injection vulnerability allows threat actors to execute arbitrary system-level commands on a server by exploiting improperly sanitized user input.

Save up to 52% off Lifelock Identity Theft Protection!

Your personal info is in endless places. And any one of them could accidentally expose you to identity theft. That's why LifeLock monitors hundreds of millions of data points a second for identity theft. LifeLock. For the threats you can't control.

Preferred partner (What does this mean?)View Deal

In this case, the bug is tracked as CVE-2023-33538 and has a severity score of 8.8/10 (high). It affects multiple models, including TP-Link TL-WR940N V2/V4, TL-WR841N V8/V10, and TL-WR740N V1/V2.

All of these models reached their EoL long ago - between 2010 and 2018. That means that they are no longer receiving updates, and that TP-Link will not be addressing the command injection vulnerability mentioned above.

Usually, when a bug is added to KEV, Federal Civilian Executive Branch (FCEB) agencies have three weeks to apply the patch. Since in this case, there is no patch, users are urged to replace old hardware with newer versions. The deadline to complete the removal is July 7, 2025.

Most OEMs advise this for all of the equipment that reached end-of-life status, both hardware, and software.

Despite being a decade old, these devices are still quite popular - as ,ost can still be purchased on Amazon, where one of the models has more than 9,000 positive reviews, and another has more than 77,000 reviews and ranks well among other similar routers.

“Users should discontinue product utilization,” CISA warned on its website.

The proof-of-concept exploits are “widely available” online, Cybernews noted, highlighting these types of flaws are most dangerous on publicly exposed routers with remote access features. It doesn’t mean they cannot be exploited within the same local network.

• This worrying botnet targets unsecure TP-Link routers - thousands of devices already hacked
• Take a look at our guide to the best authenticator app
• We've rounded up the best password managers