News

In today’s technology landscape, AI is a key catalyst driving innovation and new business models. What began as basic text and image generation has evolved into sophisticated agentic AI, autonomous systems enhanced by human oversight, delivering scalable, efficient solutions that give businesses a competitive edge.

If we look back over the decades, virtual assistants such as Amazon's Alexa and Apple's Siri were primarily designed as single-skill tools. Their functionalities were often limited to specific commands, such as playing music, setting reminders, or providing basic information.

While groundbreaking at their inception, these virtual assistants operated within clearly defined parameters, lacking the ability to integrate information across different domains or perform complex reasoning. Their utility, though significant, was circumscribed by their specialized nature.

However, the current trajectory of AI development points towards a profound shift, the emergence of autonomous agents that are now being embedded into the enterprise fabric agents. These advanced AI systems are designed to process and synthesize information from various sources, allowing them to tackle more complex assignments and engage in nuanced interactions.

This transition is not merely an incremental improvement but a fundamental redefinition of AI's potential, enabling agents to understand context, anticipate needs, and even learn from interactions to enhance their performance over time. This leap in capability allows for a more fluid and intuitive user experience, bridging the gap between isolated functions and integrated problem-solving.

The broader world of AI can be broadly categorized into two principal domains, each with distinct applications and implications:

Consumer AI: Everyday AI, like ChatGPT, found in personal devices, boosts individual productivity and convenience. However, these are largely reactive tools that require user prompts.

Enterprise AI: Business-focused AI, optimizing operations, decision-making, and automation across industries. Examples include AI for healthcare diagnostics, financial fraud detection, or manufacturing predictive maintenance. It aims to create efficiencies and competitive advantages.

The distinction between consumer and enterprise AI, while useful for categorization, is becoming increasingly blurred as AI technologies mature and become more interoperable. The advancements in natural language processing and machine learning, initially driven by consumer demand, are now finding profound applications in enterprise settings, and vice-versa.

This synergistic development is accelerating the overall progress of AI, paving the way for even more sophisticated and integrated AI agents capable of navigating the complexities of both our personal and professional lives.

Agentic AI reframes the AI landscape by moving beyond traditional consumer and enterprise applications toward autonomous decision-making systems that act with purpose and context.

It's vital to recognize these distinct verticals and manage expectations accordingly. A common pitfall in enterprise AI is the assumption that business tools will function with the same seamlessness as consumer AI.

This “expectation gap” necessitates adjusting our approach to integrating these technologies into enterprise settings. Understanding this distinction is fundamental to defining a clear roadmap for agentic AI adoption in the business world.

For effective agentic AI implementation in an enterprise context, the XDO Blueprint is highly recommended:

X (Experience): AI's primary purpose should be to enhance human experiences. This includes improving customer experience, employee experience, partner experience, and even machine-to-machine interactions within connected systems.

D (Data): Enterprises can only leverage AI effectively if they thoroughly understand and manage their data. A significant obstacle is that enterprise data is often siloed within applications. Organizations must prioritize separating data from applications, defining metadata, and structuring their data catalogues, marketplaces, and contracts efficiently.

O (Operations): This encompasses two broad areas: IT Operations: AI agents can significantly automate IT tasks, from problem detection and correction to fulfilling requests and deploying resources. They bridge the gap between humans and machine data, generating valuable insights.

Business Operations: Agentic AI can drive autonomous, intelligent operations, leading to unprecedented efficiency and agility. It can transform workflows, decision-making, and customer experiences, enabling proactive adaptation and strategic growth. Without this framework, agentic AI risks becoming merely another underutilized tool in the enterprise arsenal.

Given the regulatory and governance frameworks under which businesses operate, orchestration is critical. Unlike deterministic business processes, agentic systems are inherently probabilistic.

Companies will soon contend with a growing number of AI agents from diverse vendors, built on various technologies. The challenge extends beyond mere deployment to orchestrating these agents across the entire enterprise.

While many SaaS companies are pushing AI agents and enterprises are developing their own on hyperscaler platforms, current AI orchestration solutions often focus on managing only their proprietary agents.

The real need is for enterprise-wide orchestration, connecting disparate subsystems and ensuring AI-driven processes function seamlessly across the entire business.

Companies that adopt the XDO approach, linking experience, data, and operations, are more likely to achieve effective agentic AI implementation.

We've featured the best AI chatbot for business.

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

• A rumor claims the Steam Deck 2 may not be here until 2028
• However, this feels like tentative speculation for a couple of reasons
• It's entirely possible that we might be in for a long wait for the Steam Deck sequel, though

The Steam Deck 2 might not arrive until 2028, or that's the latest theory floating down from the rumor mill regarding the sequel to Valve's handheld.

KitGuru noticed that Kepler has again dropped another rumor via the Neogaf forums (there've been a few hardware-related nuggets this week), and it's a very short and not-so-sweet revelation that: "Steam Deck 2 is 2028".

The comment came in a post that's actually about the Asus ROG Xbox Ally, when somebody commented about buying one of those - the portable arrives in October - weighed against the prospect of Valve bringing out something better in the near(ish) future.

Kepler quickly put paid to the idea that the Steam Deck 2 might be even remotely on the horizon. If the leaker is correct, we are going to be waiting three years for the next take on Valve's gaming portable - with it arriving six years after the original Steam Deck debuted - but that is, of course, a significantly sizeable if as I'll discuss next.

Obviously, this is just a rumor, and the nature of it - a throwaway one-liner in a forum post - should give more than a brief pause for thought.

As KitGuru points out, Kepler probably didn't tap a Valve contact for this info - though we don't know that for sure, it seems likely to be the case. This is because Kepler primarily deals in GPU (or CPU) rumors and so the likelihood is that they're basing this on information drawn from an AMD roadmap - which pertains to the expected chip to be used as the engine of the next Steam Deck.

So, that makes this speculation feel a littler shakier, but that said, Kepler has proved to be one of the more reliable leakers for GPU info - so I wouldn't dismiss this nugget out of hand, by any means.

Valve is already rumored to be working on a Steam Deck 2 - which is hardly a surprise - but seemingly the development road could be a long one. Back in 2023, a year after the Steam Deck launched, the expectation was a next-gen version of Valve's handheld wouldn't arrive for a few years from then - meaning people were hoping for 2026, not 2028.

If this rumor is true, what it also indicates is that Valve is planning on making the Steam Deck 2 a huge uplift with performance, with a much beefier SoC inside - the hardware available in 2028 is going to be a lot beefier, of course. And that aligns with previous chatter suggesting the company doesn't want an incremental release with the sequel to the Steam Deck, but a big stride forward.

All that said, I would underline that we should be very cautious around this speculation for now.

• Where to buy the Asus ROG Xbox Ally handhelds: latest pre-order details, key information, and best early retailer links to bookmark now
• Best Steam Deck games 2025: top picks for Valve's handheld console
• The Razer Handheld Dock Chroma offers Steam Deck owners a premium design and, of course, plenty of RGB

Let's be honest. When movies leave your favorite streamer, some of them will be missed more than others – so when Rappin' leaves Prime Video in just under a week from now, there probably won't be people openly crying in the streets.

But while you can probably live without Mario Van Peebles showing his neighbors "how to drive out riffraff with rap music", there are some real gems leaving Prime Video at the end of this month and there are three in particular I think are must-watch movies.

For my picks this week I've tried to cover a wide range of movies, and I think it's fair to say that other than their impressive Rotten Tomatoes ratings these films don't exactly have much in common: there aren't many killer clowns in the literary biopic Capote or the urgent, suspenseful The Taking of Pelham One Two Three. But all three movies are guaranteed to entertain, albeit in very different ways.

Philip Seymour Hoffman is superb as the titular writer, and this dramatization of real events follows Truman Capote as he investigates the murder of a Kansas family. The big-city writer travels to small-town America with his friend, Harper Lee, and his research into the case and the friendship he forms with one of the killers lead to the creation of one of the classics of American literature, 1965's In Cold Blood.

The movie has a very high 89% rating from the Rotten Tomatoes critic roundup, and it comes with high praise from The New York Review of Books: "Capote is the only movie I know of that comes close to suggesting successfully what the complex process of creating a literary work actually looks like."

The Village Voice rated it too (no link available), saying: "Capote is a cool and polished hall of mirrors reflecting the ways in which Truman Capote came to write (and be written by) In Cold Blood." And Empire gave the movie the full five stars. It's "an outstanding film, anchored by a great central performance."

This is unlikely to appear in a double bill with Capote, but Damien Leone's slasher sequel has a whopping 87% rating from critics with strong stomachs. Once again Art the Clown targets teenagers in a small town, and over its two-plus hours the film delivers a stylish and genuinely frightening horror story.

This is not a movie for the faint hearted. "Skip dinner before you watch," LA Weekly recommends, "and maybe shower and then go do something nice for humanity afterward?" And HorrorBuzz was cautious in its praise, saying: "Mae West was once quoted as saying, 'Too much of a good thing can be wonderful!'. Here it depends entirely on how you felt about the first Terrifier."

But even Common Sense Media was won over. "The heinous Art the Clown returns in this intensely gory sequel that tries much harder – and is much smarter – than the original movie."

This classic thriller was remade, largely unsuccessfully, as a Denzel Washington vehicle in 2009. But the one you want is the 1974 original starring Walter Matthau, which is currently sitting with an entirely deserved 98% on Rotten Tomatoes. It's the story of an audacious crime: four men get on a New York subway train, separate the carriage and hold 17 passengers hostage. If their demands aren't met, they'll shoot a hostage every hour.

Matthau's "wonderfully weary sense of irony is perfect," says The Hollywood Reporter, while the late Roger Ebert told Chicago Sun-Times readers that "What's good about Pelham's example of the form is that the performances are allowed enough leeway so that we care about the people not the plot mechanics. And what could have been formula trash turns out to be fairly classy trash, after all."

As Empire put it, it's "the kind of gritty, relentless thriller that could only come from the ‘70s" and it's influenced lots of culture: "Quentin Tarantino would later nick the criminals using colors as codenames gambit for Reservoir Dogs; the Beastie Boys reference it in the song Sure Shot."

• A shocking lawsuit against Amazon makes me want to cancel my Prime Video subscription immediately
• I can’t wait for Fallout season 2 to dive into New Vegas – here are 6 things I’m desperate to see when the Prime Video show returns
• I’ve seen all the best Prime Video movies coming in August 2025 and this horror comedy is one you shouldn’t miss

• Passwordstate's latest version patches an authentication bypass flaw
• It could be abused to access the Passwordstate Administration section without authentication
• There are workarounds, too

Passwordstate, an enterprise-grade password manager tailored for organizations and IT and security teams, is urging users to update their instances to the newest version and mitigate risks of potential authentication bypass attacks.

“Today we have released build 9972, which includes 2 security updates,” Click Studios, the company behind Passwordstate, said in its security advisory. “We recommend customers upgrade as soon as possible.”

The changelog for Passwordstate 9.9 - Build 9972, talks about a “potential authentication bypass when using a carefully crafted URL against the core Passwordstate Products’ Emergency Access page”.

The CVE ID for the vulnerability is currently pending, so we don’t know the severity at the moment, but we do know that exploiting it allows threat actors to gain access to the Passwordstate Administration section. Depending on how easy it is to pull off, the severity score could be quite high.

Speaking to BleepingComputer, Click Studios also said there was a workaround for those who cannot patch that fast: "The only partial work around for this is to set the Emergency Access Allowed IP Address for your webserver under System Settings->Allowed IP Ranges. This is a short term partial fix and Click Studios strongly recommends that all customers upgrade to Passwordstate Build 9972 as soon as possible."

Passwordstate is a secure password vault used to store, organize, and control passwords, API keys, certificates, and other secrets. It is primarily an on-prem solution, although cloud-based options are available, as well. It is praised for its enterprise-level functionality and affordability versus higher-priced PAM tools, but also criticized for its steeper technical learning curve, setup, server requirements, and UI complexity.

Click Studios claims it is used by more than 370,000 users working in 29,000 companies, including government agencies, financial institutions, global enterprises, Fortune 500 companies, and others.

Via BleepingComputer

• Hackers are distributing a cracked password manager that steals data, deploys ransomware
• Take a look at our guide to the best authenticator app
• We've rounded up the best password managers