Ruark Audio is turning 40, and rather than celebrate with a badge slapped on a press release and a commemorative mug no one asked for, the British manufacturer has introduced the R410 Anniversary Edition, a limited production version of its all in one music system finished in White Oak veneer with an ebonised grille and ebony veneer inlay.

Only 500 systems will be produced worldwide, each carrying an anniversary badge. The system is listed at £1,399, with wider European pricing reported at €1,599, and Ruark says the Anniversary Edition is due for release in August. Fidelity Imports has not yet confirmed U.S. pricing, but we will update this story when that information becomes available.

That makes this less of a new platform and more of a carefully dressed version of a product Ruark already understands rather well. The R410 remains part of the company’s 100 Series, sitting alongside the R610 Music Console, R710 CD Hi-Fi Console, and R810 Radiogram. The difference is that the R410 keeps the entire system in one cabinet: streamer, amplifier, DAC, phono input, radio, HDMI ARC/eARC, speakers, display, and physical controls. No speaker matching. No rack. No cable spaghetti slowly turning your living room into a failed Maplin or Crazy Eddie’s demonstration.

The Anniversary Edition’s biggest visual change is the cabinet. Ruark has moved from the standard R410 finishes to a White Oak veneer cabinet, paired with an ebonised grille and ebony veneer inlay. That contrast matters. The regular R410 already leaned into mid century modern cues, but this version looks more deliberate and less like another lifestyle audio box trying to win a design award by being beige and harmless.

The matching limited edition R-CD100 CD player, finished in ebonised casework, will also be available alongside it. That is the right move for anyone with shelves full of discs they actually intend to play. Having already reviewed the R-CD100 CD transport, it is also a logical add-on for any compatible Ruark system.

Related Reviews:

Connectivity Galore? 

Ruark’s pitch is clear: one box that can handle streaming, radio, vinyl, TV audio, local files, Bluetooth, and optional CD playback. The R410 Anniversary Edition supports Apple AirPlay 2, Google Cast, Spotify Connect, TIDAL Connect, and Qobuz Connect. It also includes Internet Radio, FM with RDS, UPnP/DLNA media server support, Bluetooth 5.1 with aptX HD, AAC, and SBC, plus Wi-Fi networking listed as 802.11 a/b/g/n/ac/ax.

DAB/DAB+ support is also included, although that matters far more in the U.K., Europe, Australia, and other active DAB markets than it does in North America, where HD Radio remains the relevant terrestrial digital radio format.

The wired side is not an afterthought either. There is HDMI ARC/eARC for TV audio, an optical input up to 24-bit/96kHz PCM, stereo RCA line input, a moving magnet phono input with adjustable gain, Ethernet, USB-C file playback and charging, and a mono RCA subwoofer output.

That is the reason the R410 is more interesting than most wireless speakers. It acknowledges that people still own televisions, turntables, CD collections, NAS libraries, and occasionally brains. Unless you live in Maine at the moment and still think Platner was a viable candidate for the U.S. Senate.

Inside, the R410 Anniversary Edition uses a fully active 120 watt Class D amplifier, twin bass reflex cabinet architecture, two 100mm Ruark NS+ bass mid woofers, and two 20mm silk dome tweeters. Ruark also specifies Burr-Brown 32-bit/192kHz DAC and ADC stages, adjustable bass and treble, and switchable Stereo+. Hi-res file support goes up to 32-bit/192kHz for FLAC, AIFF, ALAC, and WAV. MP3 is supported up to 48kHz/320kbps, while AAC is supported up to 96kHz/320kbps. 

The display and control system are part of the appeal. The R410 uses a 4-inch colour TFT display with auto dimming and Ruark’s familiar RotoDial control, supported by a rechargeable wireless remote. That sounds like a small thing until you have spent ten minutes inside a rival app wondering why changing inputs now feels like filing a planning application with the Long Branch council. Ruark’s best products work because they remain physical, tactile, and understandable. The company has figured out that convenience should not mean surrendering every useful control to a phone.

We have spent a fair bit of time with Ruark over the past two years, and the appeal has become clearer with each product. The company is not chasing the usual wireless-speaker race to the bottom, nor is it pretending that every listener wants a rack full of separates. 

The R410 Anniversary Edition lands in the middle of that strategy. It is not as ambitious as the R810. It is not as flexible as the R610 or R710 if you want to choose your own loudspeakers. But it may be the cleanest expression of what Ruark is trying to do for listeners who want better sound without turning their living room into a dealer demo room. It is for someone who wants one system, proper source flexibility, attractive industrial design, and enough sonic ambition to make a Sonos or soundbar solution feel rather underdressed.

Key Specifications:

• Model: Ruark R410 Anniversary Edition
• Finish: White Oak veneer cabinet with ebonised grille and ebony veneer inlay
• Production: Limited to 500 systems worldwide
• Price: £1,399; reported European price €1,599
• Availability: Due August
• System type: All in one active music system
• Amplification: Fully active Class D, 120W total
• Speaker drivers: 2 × 20mm Ruark silk dome tweeters; 2 × 100mm Ruark NS+ bass mid woofers
• Cabinet: Twin bass reflex
• Frequency response: 40Hz to 22kHz, typical in room
• DAC: Burr-Brown 32-bit/192kHz
• ADC: Burr-Brown 32-bit/192kHz
• Tone controls: Adjustable bass and treble, ±6dB
• Sound processing: Switchable Stereo+
• Display: 4-inch colour TFT with auto dimming
• Controls: RotoDial, rechargeable wireless remote, 20 global presets
• Streaming: Spotify Connect, TIDAL Connect, Qobuz Connect
• Wireless platforms: Apple AirPlay 2, Google Cast
• Bluetooth: Version 5.1 with aptX HD, SBC, AAC
• Network playback: UPnP/DLNA media server compatible
• Wi-Fi: 802.11 a/b/g/n/ac/ax
• Ethernet: RJ45, 10/100 Mbps
• Radio: Internet Radio, DAB/DAB+, FM with RDS
• HDMI: ARC and eARC
• Optical input: Up to 24-bit/96kHz PCM
• Line input: Stereo RCA, up to 2.3Vrms
• Turntable input: RCA moving magnet phono stage with adjustable gain, 2 to 7mV
• Subwoofer output: Mono RCA, 2.0Vrms
• USB-C: File playback and 5V/5W charging
• Hi-res file support: FLAC, AIFF, ALAC, WAV up to 32-bit/192kHz
• Compressed file support: MP3 up to 48kHz/320kbps; AAC up to 96kHz/320kbps
• Dimensions: 132 × 560 × 290mm cabinet; 150 × 560 × 325mm maximum including feet, controls, and cables
• Weight: 9.5kg
• Power consumption: 2W standby, 8W typical
• Included: R410, 2m AC power cable, quick start guide, telescopic aerial and spanner, rechargeable wireless remote

The Bottom Line

The Ruark R410 Anniversary Edition is not trying to reinvent the R410, and that is probably wise. The core system already made sense: serious streaming support, a real phono input, HDMI ARC/eARC, radio, optional CD playback, useful physical controls, and a cabinet that does not look like it escaped from a router factory.

What makes this version different is the execution. The White Oak veneer, ebonised grille, ebony inlay, anniversary badge, and 500 unit production run push it closer to collectible territory without turning it into a ridiculous luxury object. For listeners who want a handsome all in one system that handles modern streaming, vinyl, TV audio, radio, and CDs without demanding a full rack, this is Ruark making a very clear argument.

For more information: ruarkaudio.com

Related Reading:

The iPad Pro with M6 is expected in early 2027 and Apple may release upgraded Apple Pencil models with easier-to-replace batteries in time for new EU regulations.

The European Union mandated in 2023 that consumer electronics must have easily replaced batteries by 2027. One of Apple’s notoriously impossible-to-repair products is the Apple Pencil, which could see some design changes to meet this mandate.

According to the “Power On” newsletter from Bloomberg, Apple is expected to release an updated Apple Pencil Pro and Apple Pencil with USB-C in early 2027. That timing aligns with previous reports of an iPad Pro refresh during the same release window.

No details about the new Apple Pencil models were provided. There isn’t any information about what new features might be introduced or how the design might change to accommodate easily-replaced batteries either.

Due to the regulatory requirements, the entire upgrade could simply be focused around the battery change. Currently, the Apple Pencil is a hunk of plastic filled with glue and is virtually impossible to repair.

Rethinking Apple Pencil design

Apple’s focus on selling what is essentially a plastic unibody stylus means it had no seams, screws, or separation points anywhere. Adding the USB-C port in the Apple Pencil with USB-C that hides behind a sliding mechanism was quite the design change on its own.

Because of the sliding mechanism, the lower-priced Apple Pencil with USB-C may be the easier product to redesign with replaceable batteries in mind. It isn’t clear how Apple might adapt the solid Apple Pencil Pro for the regulation.

To offer a guess, since the Apple Pencil tip is replaceable and offers the single seam in the device, it could also act as an entryway for internal access. Of course, the use of glue would need to be reduced or thrown out entirely.

Outside the iPhone, which has already been designed to accommodate the rules, EU battery regulations will require dramatic changes to some of Apple’s product designs, or necessitate quite clever solutions. It remains to be seen how something like AirPods might meet the regulatory standard.

Enterprise AI teams are giving agents more freedom at the same moment their confidence in automated testing is collapsing.

Half of enterprises have deployed an AI agent or LLM feature that passed internal evaluations and yet still caused a customer-facing failure — one in four more than once — according to the June 2026 VB Pulse survey of 157 qualified enterprise respondents at companies with 100 or more employees.

The sample is self-selected rather than a probability sample, so the findings should be read as directional, not precise.

But enterprises are not responding by slowing automation: 66% of respondents already permit some production deployment without human review or are building systems intended to do so within the next 12 months. Only 5% say they fully trust the automated evaluations that would make those release decisions.

That mismatch is the evaluation gap: the autonomy ceiling is rising faster than the assurance beneath it. 

It also fits a broader thesis that will be explored at VB Transform 2026: enterprises ship agents first, while the control layers around identity, evaluation, cost, context and orchestration are arriving later. The next year will be a retrofit cycle, with buyers shifting budget toward the systems that make agentic deployments governable and dependable.

Why a passing evaluation is not a working agent

Traditional software testing usually asks whether a defined input produces an expected output. Agent testing is harder because the system may choose its own sequence of steps, call tools, retrieve data, alter state and respond differently from one run to the next.

An agent can make several individually plausible decisions and still reach the wrong result. It may retrieve the correct account but update the wrong field. It may draft a valid refund request but send it without approval. It may call five tools successfully before a sixth step leaks sensitive information or leaves a workflow incomplete.

The survey shows enterprises already recognize this limitation. The most common reason for distrusting automated evaluation is poor alignment with real-world outcomes, cited by 29% of respondents. Bias or inconsistency follows at 21%, lack of explainability at 18%, and data leakage or privacy concerns at 17%.

That hierarchy matters. Enterprises are saying the score often does not predict what happens when a customer, employee or business process encounters the agent in production — not that automated scoring is too slow or expensive.

NIST makes a similar point in its Generative AI Profile: measurements gathered in controlled environments may not transfer cleanly to deployment because behavior changes with prompts, users, context and operating conditions. Its guidance calls for field testing, post-deployment monitoring and clear processes for escalating failures.

Capability is not consistency

A single successful run proves that an agent can complete a task. It does not prove that it will complete the task reliably.

Anthropic’s guidance on agent evaluation distinguishes between measuring whether a system succeeds at least once across repeated attempts and whether it succeeds every time. That distinction is essential for customer-facing or operational workflows. A model that occasionally produces an excellent answer may still be unacceptable if the same task fails unpredictably on the next attempt.

Enterprise teams should therefore treat repeatability as a first-class metric. That means running the same scenario multiple times, varying phrasing and context, testing tool failures, and measuring whether the final business outcome remains correct even when the route changes.

The evaluation set also has to evolve. Every production incident should become a permanent regression test. Customer escalations, failed tool calls, incorrect approvals and data-handling mistakes should feed back into the pre-deployment suite rather than remaining isolated support cases.

Autonomy should expand by risk, not by ambition

The survey does not imply that every agent action should require a person. Human review cannot scale across millions of low-consequence decisions.

But zero-human operation should be earned by demonstrated reliability and bounded by the consequences of failure.

Low-risk actions such as drafting internal summaries or categorizing documents can tolerate broader autonomy. Financial transactions, customer communications, code deployment, access-control changes and data deletion need stricter thresholds, repeated consistency tests, policy checks, rollback mechanisms and clear human escalation paths.

The risk isn’t evenly distributed by company size, either. Larger enterprises — those with 2,500 or more employees — are moving toward zero-human deployment fastest, at 70% versus 64% for smaller companies, and they’re also shipping more agents that go on to fail a customer, at 54% versus 48%. 

That is the warning for enterprise leaders. Removing the human from the loop does not remove uncertainty. Without stronger assurance, it converts uncertainty into an automated production decision.

The market will keep pushing toward greater autonomy because the economic incentive is real. The organizations best positioned won’t be those that remove people fastest — they’ll be the ones that treat repeatability and regression testing as seriously as deployment speed.

Enterprise companies are running AI agents ahead of the controls needed to manage them — and they deployed that way knowingly. That is the central finding from VentureBeat Research’s June survey of 573 technical leaders at companies with 100 or more employees, fielded across five parallel surveys of the agentic stack. 

Enterprises are now retrofitting to catch up with their own standards, and they are budgeting for it: Roughly six in 10 enterprises plan to switch or add vendors in each of five control layers within the next 12 months, and roughly a third — depending on the layer — plan to move within the quarter, the research finds.

There are five main layers where enterprises are building: identity for agents (which agent is allowed to do what, under whose credentials); evaluation of agent output (whether the work is any good); cost telemetry (what each agent costs to run); the context layer (the business data and definitions agents draw on to answer); and the orchestration control plane (the software that coordinates multi-step agent work).

Enterprises are already paying the price for deploying agents ahead of adequate control functions. Fifty-four percent of companies had an agent security incident or near-miss caught before harm in the past 12 months. Twenty-seven percent exercise only reactive control of agent spend — they learn what an agent costs when the invoice arrives, with no per-agent budget or ceiling in place.

Here are the five findings that anchor the set — one finding per layer of the tech stack — and what the data suggests doing first in each.

Expensive hardware is idle: 86% of GPU operators report utilization of 50% or less

Eighty-six percent of enterprises that run their own GPUs report utilization of 50% or less. Wall Street has spent the quarter debating whether the AI buildout is overbuilt. This is buy-side measurement, from the enterprises doing the buying, and the research says the most expensive hardware in buildings of these enterprises runs at no more than half its capacity.

The measurement gap compounds it: A minority 44% rigorously track what their AI compute actually costs and returns. Everyone else is only estimating. And the enterprise shopping process continues regardless: 45% of these enterprises say the emerging compute option they are most likely to evaluate in the next 12 months is an AI-specialized cloud (CoreWeave, Lambda, Crusoe, Nebius). However, under 2% of these enterprises report using one of these neoclouds today.

Moreover, roughly one in three companies appears to be considering a hedge against Nvidia: Asked which emerging compute option they are most likely to evaluate in the next 12 months, 32% of enterprises named non-Nvidia accelerators (AWS Trainium, Google TPUs, AMD), while 28% named next-generation Nvidia GPUs. The data suggests that enterprises should measure the utilization and per-workload cost of the GPUs they already own before committing budget to new compute — whether that’s an AI-specialized cloud contract, new accelerators, or more GPUs. 

Most deployed “agents” do single-prompt work: 71% say a quarter or fewer complete multi-step tasks on their own

Seventy-one percent of enterprises say a quarter or fewer of their deployed “agents” can complete multi-step work on their own; the rest are single-prompt chatbots. Only 10% say true agents are the majority of what they run. To be sure, the respondents reported that they are in a position to know these things: 81% said they recommend or decide AI purchases at their companies.

That finding — that most agents are actually just chatbots in trenchcoats — lands amid adoption claims across the industry running well ahead of what enterprises are actually running. Gartner predicted 40% of enterprise applications will be integrated with task-specific AI agents by the end of 2026, up from less than 5% in 2025. It also warned that the most common misconception is referring to these AI assistants as agents, a misunderstanding known as “agentwashing.”

Meanwhile, Zapier’s enterprise survey said 72% reported deploying or testing autonomous agents; and Writer’s 2026 survey has 97% of executives saying their company deployed AI agents in the past year. 

Those surveys asked whether companies have deployed something called an AI agent, and companies said yes. Our survey asked the people running those deployments a harder question: Of the agents you have in production, how many can complete a multi-step task without a person driving each step? The gap matters for two practical reasons. First, the inflated adoption figures are the benchmark boards and vendors use to pressure technical leaders into moving faster — and this data says the real bar is far lower than the headlines suggest. Second, the label determines the bill: A single-prompt chatbot with a human reading every answer needs none of the identity, evaluation, and cost controls this report covers, while a true multi-step agent needs all of them. 

66% let agents push to production on automated evals alone — or are engineering toward it. 5% fully trust those evals

Two-thirds of enterprises fall into one of two camps: 34% already allow an AI agent to push a code or system change to production based on automated evaluation results alone, with no human reviewing it, and another 33% are actively engineering their pipelines to allow that within the next 12 months. Only five percent fully trust the automated evaluations that would make that decision.

The distrust is earned. Half of enterprises shipped an agent that passed internal evaluations and then caused a customer-facing failure in the past year; a quarter watched it happen more than once. Asked to name the biggest weakness in their current evaluations, more enterprises chose “poor alignment with real-world outcomes” than any other answer — 29% of respondents.

And most of the checking happens before an agent ships, then stops. Once agents are live with real users, only 23% of enterprises run real-time quality checks on the answers those agents produce. Another 51% monitor system health only — uptime, request traces, and gateway logs — which tells them the agent is running, and nothing about whether its answers are right. The first move: Before removing human review from any workflow, test your evaluations against production outcomes rather than internal benchmarks, and instrument answer quality, not just uptime.

This finding is explored in more depth in VentureBeat’s related coverage of the evaluation gap, which found that larger enterprises are moving faster toward zero-human deployment while also failing more often — and outlines a regression-testing framework built on production outcomes rather than internal benchmarks.

69% run credential sharing somewhere in the agent fleet — and those companies get hit far more often

Sixty-nine percent of companies allow agent credential sharing somewhere in their agent fleet during runtime – meaning multiple agents operating under one API key or service account. Those companies were far more likely to get hit: Organizations with credential sharing anywhere in the fleet experienced a security incident or near-miss at a 63.5% rate (47 of 74), against 40.9% (9 of 22) where every agent has its own scoped identity. 

The takeaway for enterprises is this: Give every agent its own scoped identity, starting with the agents that touch production systems.

57% traced a confident, wrong agent answer to their own missing or inconsistent business context

Fifty-seven percent of enterprises traced at least one confident, wrong agent answer in the past six months to missing or inconsistent business context: wrong metrics, stale definitions, absent documents. Most of them watched it happen more than once.

Most enterprise companies are fixing this, even though they’ve moved forward with agent deployment already: 25% already run a governed semantic layer, or one governed definition of the business that every AI reads from, in production. However, 34% are still building one, and 41% haven’t started. The takeaway: Govern the definitions your agents answer from, metrics and entities first, before scaling the agents that depend on them.

The quarter where agent technology “portability” became a priority

One more shift is worth reporting with its limits stated plainly. In our spring orchestration survey wave, the top concern about provider-controlled orchestration was security and permissioning limits (32%). By June, vendor lock-in led at roughly a third, with security limits at 28%. 

Those are two snapshots one quarter apart, and here’s one possible explanation for why portability became a top issue for enterprises. Our June survey went into market after a June 12 U.S. Commerce Department export order took Anthropic’s Claude Fable 5 offline for enterprises for roughly three weeks. Meanwhile, Chinese company Z.ai released GLM-5.2’s open weights under an MIT license on June 16 at roughly one-sixth of GPT-5.5’s price; and Tencent’s Hy3 arrived July 6 under Apache 2.0; and OpenAI previewed GPT-5.6 on June 26 to a small group of government-vetted partners, opening it broadly on July 9 after the government’s review cleared. The open-weight releases in particular promise enterprises more control over their agents, and while we haven’t established a causal link here, the timing is worth noting.

The posture data matches the mood: 51% now expect their primary control plane for enterprise agents to be hybrid — provider-native plus external orchestration — by the end of 2026, up from 34% in the spring survey wave. Enterprises reporting that they rely purely on provider-managed agent services fell from 12% to 7%.

Five layers, no incumbents, 12 months

The synthesis across all five surveys reveals a huge “buying” window. In each of the five control layers, 57% to 64% of enterprises plan to switch or add vendors within 12 months — 64% in infrastructure and in evaluations, 59% in agent security, 57% in retrieval and context — and 26% to 38%, depending on the layer, plan to move within a quarter. No layer has an established incumbent: The most common evaluation tooling is the model provider’s built-in evals, tied with no dedicated tooling at all (17% each); 82% of respondents name provider-native or hyperscaler controls as their primary agent security layer; and provider-native retrieval leads the context technology layer (RAG, etc) as well. 

Most enterprises are defaulting today to the built-in tools that ship with the big AI platforms they already use: Anthropic, OpenAI, Google, Microsoft, and AWS. That holds true across every one of these agentic technology layers: enterprises are looking to their primary cloud and model providers to supply the guardrails, evaluations, and retrieval solutions already bundled into those providers’ offerings.

Those defaults are winning on convenience, and they’re also what the coming spending decisions will test. The survey didn’t ask which direction that money moves — toward the platforms’ built-in tools or toward the specialists challenging them — which is exactly why every contract in these five layers is worth watching over the next four quarters.

The Q3 survey wave will measure whether the enterprises made good on these budget plans: whether their agents gained scoped identities, whether evaluations got tested against production outcomes, whether GPU utilization rose, and whether the semantic layers under construction shipped.

VentureBeat will release the full Q2 reports across all five VB Pulse trackers at VB Transform, July 14–15 at Hotel Nia in Menlo Park, where we convene enterprise technical leaders building autonomous agents in production. 

Disclosure: VentureBeat produces both this research and VB Transform

OpenAI on Thursday launched ChatGPT Work, a new AI agent embedded inside its flagship chatbot that aims to transform ChatGPT from a question-and-answer tool into an autonomous work platform capable of executing complex, multi-step tasks across users’ email, calendars, code repositories, and messaging apps.

The product is powered by OpenAI’s latest flagship model, GPT-5.6, and is designed to go far beyond generating text. ChatGPT Work can gather context from connected apps, files, and workflows to produce finished documents, spreadsheets, presentations, reports, and websites. The agent takes a stated outcome, breaks it into smaller steps, and stays with complex projects for hours, completing them independently.

The launch marks OpenAI’s clearest attempt yet to reposition ChatGPT as a workplace platform rather than a chatbot — and it arrives at a moment of extraordinary financial significance for the company. Last month, OpenAI confidentially submitted a draft S-1 registration statement to the SEC, initiating what could become one of the largest technology IPOs in history, with reported valuations clustering between $730 billion and $852 billion and annualized revenue that has blown past $25 billion.

In a short demonstration and conversation with VentureBeat on Friday, Ty Geri, a product manager at OpenAI who helped build ChatGPT Work, said the product’s mission is to democratize the kind of agentic AI capabilities that OpenAI’s internal engineering tool, Codex, has already demonstrated. “What’s really exciting is we’ve seen how much Codex has been able to push the frontier of what we can get done with these AI tools, as opposed to just getting information or answers or guidance,” Geri said. “Our internal adoption of Codex is literally an exponential curve across every single product function and every single use case.”

Why OpenAI built a persistent virtual machine that works from the beach

The core architectural bet behind ChatGPT Work is a persistent cloud-based virtual machine that runs on OpenAI’s servers, always available to the user regardless of which device they happen to be on. That marks a deliberate departure from competitors whose agents require a local machine to remain powered on and connected.

“What’s really exciting about ChatGPT Work is that it’s a virtual machine in the cloud that’s always on for you, and this is available across all of our paid tiers,” Geri said. “All Plus users are getting this. I think that’s a very unique aspect of this.”

The mobile-first aspect of the launch is something Geri described as “missing from the market.” He pointed to the ability to create a website on a phone and share it with collaborators as a particularly novel capability. “Sites are new in general to Codex. They launched in Codex about a week and a half ago, but now we’re launching also in web and mobile. You can create a site on your phone at the beach and share it with your friends,” he said.

ChatGPT Work will roll out beginning with Pro, Enterprise, and Edu users, and will expand to Plus and Business users over the next few days. In the interview, Geri emphasized that the availability of the product to Plus subscribers — not just premium tiers — is central to OpenAI’s strategy. “It’s accessible to all paid plans, including Plus users, which in my opinion is a really big feat, and really part of that OpenAI mission, which is about bringing all this power to as many people,” he said.

How MCP plugins connect ChatGPT Work to Slack, Gmail, and GitHub

The product relies on MCP-based plugins to connect to external services like Gmail, Google Calendar, Slack, and GitHub. When asked whether the plugin architecture is based on the Model Context Protocol standard, Geri confirmed: “These are all based on MCP.” He added that connecting multiple Gmail accounts — a frequent user request — “is definitely on the roadmap.”

The experience is designed to be action-oriented from the first interaction. ChatGPT Work offers a personalized onboarding flow that surfaces different suggested use cases depending on the user’s role. Geri demonstrated how the system, detecting his role as a product manager, immediately suggested tasks like evaluating AI systems, building research artifacts, and managing his calendar. “You can start with a simple task like catch me up on Slack or Teams or read today’s calendar,” Geri said. He described a scenario where the system reviewed his calendar, identified scheduling conflicts, flagged meetings requiring preparation, and then — on his instruction — declined, accepted, or rescheduled events directly.

Users can also customize the agent by teaching it their writing style, organizing outputs into projects, and — in a lighter touch — choosing a virtual pet that accompanies them in the interface. The interface also introduces a hosted website feature that allows users to build and share interactive sites directly through ChatGPT Work, turning what would typically be a static slide deck into a dynamic, collaborative artifact. “Now we suddenly have a collaborative interface that’s actually more exciting and more accessible than a slide deck, which has all these formatting restrictions,” Geri said.

Scheduling 10 bug bashes at once: what agentic productivity looks like in practice

Geri’s own usage of ChatGPT Work illustrates the breadth of tasks the system can handle. In the run-up to the product’s launch, he needed to organize pre-release testing sessions — known internally as “bug bashes” — across dozens of features and team members.

“I just come to ChatGPT Work and say, ‘Set up a bug bash for all the distinct features in ChatGPT Work. Add all the people that worked on that feature,’ and it can check Slack, it can check GitHub, it can check Docs, and find a time that works for the four highest contributors to that feature,” Geri said. “It went and scheduled 10 bug bashes, all coordinated across all those different people. That would have taken me 30 minutes at least.”

But Geri pushed back against the characterization that ChatGPT Work is limited to rote administrative work. He described using it for analytically complex tasks like identifying the biggest causes of user churn for specific product features and generating product solutions — work he said would previously have taken months. “Things that we would have spent three months doing, we can now spend a week doing — and do much more, and make a much better product,” Geri said. “Bugs that we would have found three or four weeks from now, we can now find within two days and fix for our users.”

He also described handing off the tedium of product testing itself. “It used to be that even though like the most interesting part of my job is like what to test, I would actually end up having to spend most of my job doing the testing, which is like me taking a mouse and like clicking on the same thing over and over again, like five times,” Geri said. “Instead, now I can define what do we want to test, and ChatGPT Work or Codex can actually go test it for me, deliver me that bug report, and then we can work on fixing that bug.”

What OpenAI says about data privacy when AI reads your Slack and email

When pressed on data privacy concerns — given that ChatGPT Work pulls sensitive information from workplace tools like Slack, Google Drive, and email — Geri said privacy “is incredibly important, and the most important part of this is it’s always in the user’s control.”

He pointed to OpenAI’s existing enterprise security infrastructure, noting that “enterprise accounts have ZDR, and users can always opt out of letting their conversations help improve future models, which many users do.” The comment aligns with assurances OpenAI made when it first launched ChatGPT Enterprise in August 2023, when the company wrote in a blog post that it does “not train on your business data or conversations.”

The privacy question carries additional weight now because of the sheer volume of sensitive workplace data ChatGPT Work is designed to access. Unlike a chatbot session where a user voluntarily pastes text into a prompt, ChatGPT Work actively reaches into connected systems — reading Slack messages, scanning calendar invitations, pulling GitHub commit histories — to assemble context for its tasks. That represents a fundamentally different data surface area than anything OpenAI has offered before, and one that enterprise security teams will scrutinize carefully before granting access.

ChatGPT Work enters a three-way arms race with Anthropic and Microsoft

ChatGPT Work lands squarely in the middle of what has become the defining competitive battlefield in enterprise AI: the race to build autonomous workplace agents that can go beyond generating text and actually execute tasks.

The product arrives months after Anthropic took Claude Cowork out of preview and into general availability in April, bringing its AI agent to web and mobile platforms aimed at helping enterprise users monitor and manage long-running AI-driven tasks from anywhere. Meanwhile, Microsoft made Copilot Cowork generally available worldwide on June 16, built in partnership with Anthropic to move beyond chat and into execution. The three products — ChatGPT Work, Claude Cowork, and Microsoft Copilot Cowork — now compete directly for the attention of enterprise IT departments and individual knowledge workers alike.

The convergence is striking. All three products share a remarkably similar vision: a persistent AI agent running in the cloud that can break complex tasks into steps, connect to workplace tools via plugins, and produce finished outputs rather than just conversational replies. All three work across desktop, web, and mobile.

What distinguishes OpenAI’s approach is its raw consumer distribution advantage. ChatGPT has reached 900 million weekly active users, and OpenAI now has 50 million paying subscribers. More than 9 million paying business users rely on ChatGPT for work, and 92% of Fortune 500 companies now use ChatGPT. By making ChatGPT Work available to Plus subscribers at $20 a month — not just Enterprise or Pro customers — OpenAI is betting that broad accessibility will drive adoption faster than any competitor can match.

OpenAI’s product manager says AI is a partner, not a replacement — with a caveat

When asked about the potential impact on the labor market, Geri was careful with his framing. He declined to speak broadly about workforce disruption but offered his personal experience as a product manager whose day-to-day work has been substantially reshaped by the tool.

“My job is not to schedule bug bashes and find out who contributed to a specific feature. That’s a task I do in my job, but that’s not my job,” Geri said. “My job is to make an amazing product.” He described ChatGPT Work as “a partner” and “an extension of me, certainly not a replacement,” adding: “Everybody feels far more productive than before, but is also almost working harder than before, because you get to work on all the things you want to work on as opposed to the drudgery around it.”

But Geri was also careful not to minimize the sophistication of the work the agent can handle. “I also don’t want to say that it’s only doing mundane tasks because, like something like hill climbing retention curves on a given feature is not mundane. It’s actually really hard to do,” he said. The distinction matters. If ChatGPT Work were merely automating calendar invitations and expense reports, it would be a convenience tool. The fact that Geri describes it compressing three months of analytical product work into a single week suggests something with far greater implications for how teams are structured and staffed.

An IPO-bound company needs ChatGPT Work to prove enterprise AI can generate revenue

The timing of ChatGPT Work’s launch is impossible to separate from OpenAI’s IPO trajectory. The company needs to demonstrate that it can convert its massive consumer user base into durable enterprise revenue — a narrative that becomes significantly more compelling with a product explicitly designed around professional workflows.

OpenAI said it is generating $2 billion in revenue per month, growing four times faster than Alphabet and Meta did at comparable stages, with enterprise now making up more than 40% of revenue and on track to reach parity with consumer by the end of 2026. But OpenAI remains heavily loss-making, and the company does not expect to reach profitability until around 2030, with internal projections suggesting losses of $14 billion in 2026 alone.

The competitive dynamics are unprecedented. Anthropic filed for its own IPO on June 1 at a $965 billion valuation, setting up simultaneous public listings from the two most prominent AI startups in history. Whether both can sustain their lofty valuations under the scrutiny of public market investors will depend in large part on whether products like ChatGPT Work and Claude Cowork deliver measurable productivity gains to paying enterprise customers.

The launch also caps a product trajectory that began with ChatGPT Enterprise in August 2023, accelerated through the release of OpenAI’s Operator agent in January 2025, and continued through Operator’s deprecation and shutdown on August 31, 2025, when its capabilities were folded into the ChatGPT agent framework. ChatGPT Work is the consolidation of those efforts into a single, unified product — one that pairs GPT-5.6’s three model variants (Sol for power, Luna for speed, and Terra for balanced everyday use) with a persistent cloud environment and an expanding library of MCP plugins.

The future of work may already be running in the cloud

When asked whether ChatGPT Work signals a shift toward a new kind of operating system — one where users interact with their computers primarily through an AI agent rather than through traditional mouse-and-keyboard interfaces — Geri stopped short of making sweeping predictions. But he hinted at the direction OpenAI sees ahead.

“Anybody who has worked with Codex or now ChatGPT Work will realize how exciting it is to interact with your environment and your computer via the agent,” he said. “Especially in the desktop app, where the model has access to your entire machine and can interact with websites on your behalf — it’s really able to be an extension of you and a real partner, and that certainly feels like the future.”

At the end of the interview, Geri circled back to something personal. “I’ve never enjoyed work as much as I have in the last month using ChatGPT Work and Codex,” he said — a striking admission from a product manager who, until recently, spent a meaningful share of his days clicking through the same interface five times in a row just to see if it would break. OpenAI is now asking 900 million users to believe that feeling scales. For a company weeks away from one of the largest public offerings in history, the answer to that question is worth roughly $850 billion.

Slopsquatting represents an emerging supply chain threat made possible by AI hallucinations. As developers increasingly rely on AI coding assistants, they unknowingly grant cybercriminals access to their software from day one. 

Understanding what slopsquatting is

Slopsquatting is a new type of supply chain attack that uses large language model (LLM) hallucinations to inject malicious code into development workflows. The term combines “AI slop” and “typosquatting,” a deceptive practice where attackers register misspelled or lookalike versions of popular domains to prey on users who enter URLs incorrectly.

This novel attack vector exploits LLMs’ tendency to generate fictitious software package names, which threat actors can then register and populate with malicious code.

During AI-assisted coding, the model may generate fake open-source packages — bundled collections of files, programs and installation tools. This alone is not necessarily harmful. However, if an attacker registers that fake package name, they can inject malware that gets incorporated directly into a developer’s codebase.

How AI creates a supply chain risk

Traditionally, AI safety risks stem from hallucinations, which can adversely affect users who treat misinformation as valid. However, those same hallucinations have evolved into exploitable security vulnerabilities.

Typosquatting is a deceptive practice where a cybercriminal registers a mispelled version of a popular package to trick developers. It has existed for decades, so registries have built protections against it. 

However, AI has changed the threat model. It recommends fictitious packages that sound plausible rather than making simple misspellings. Once attackers learn which hallucinated packages models tend to invent, they can register malware-filled packages under those names.

Since the hallucinated packages are not simply typoed versions of popular libraries, there are no protections against this practice at scale. For example, the registry protects against an attacker publishing “crossenv,” a squat of the popular “cross-env” package. However, it would not identify “mpn install cross-env file” or “cross-env-extended” as threats.

Hallucinations are persistent and severe

Even if many LLMs recommend the same hallucinated package, widespread compromise is still possible. Malicious packages could remain undetected in production for months or even years, allowing threat actors to passively inject malware across countless environments. 

One research team analyzed 31,267 vulnerabilities belonging to 14,675 packages across 10 programming languages. They discovered that reported vulnerabilities are increasing at an annual rate of 98%, faster growth than the 25% annual increase in the number of open-source software packages. The team also observed an 85% increase in the average lifespan of vulnerabilities, indicating a decline in security.

Real-world dangers of AI hallucinations

Malicious actors can create open-access packages under the same name as commonly hallucinated libraries. Instead of standard code, they are filled with malware. The models believe they are referring to existing packages, so they often repeat the same hallucinated names. Since the hallucinations are not random, attackers could theoretically register packages that trick tens of thousands of developers.

These packages appear legitimate. String similarity to real libraries makes them recognizable. One-character typos suggest simple mistakes rather than malicious intent. Even fully fabricated names remain believable when the AI presents them in proper context. Detection is challenging, as developers trust their coding assistants to recommend valid dependencies.

Why are LLMs hallucinating packages?

LLMs generate the statistically most likely answer rather than prioritizing accuracy. Hallucinations are relatively common as a result. One study found hallucination rates range from 50% to 82%, depending on the model and prompting method. Even GPT-4o, the best-performing model, goes no lower than 23%, even with prompt-based mitigation.

Adversarial hallucination attacks could worsen this problem. Threat actors can leverage token-level manipulation or retrieval poisoning to force models to hallucinate in ways they want, increasing the likelihood that models recommend their malicious packages.

Which LLMs are prone to slopsquatting?

While all LLMs are prone to slopsquatting, some are more vulnerable than others. The likelihood of producing hallucinated packages during code generation depends on the model. Proprietary models are four times less likely to generate hallucinated packages than open-source models.

One research group proved this by conducting 30 tests across 30 different systems. Out of the 576,000 code samples and 2.23 million packages it produced, 19.7% were hallucinations. GPT-4.0 Turbo had a hallucination rate of 3.59%, while DeepSeek 1B, the best-performing open-source model, reached 13.63%.

This research suggests that organizations relying on open-source AI tools for code generation are roughly four times more exposed to slopsquatting attacks. That doesn’t necessarily mean proprietary tools will always remain safer, though. Once attackers realize this disparity, they may manipulate proprietary LLMs to take advantage of perceived safety.

Vibe coding contributes to the problem

Software developers who use AI tools estimate that over 40 percent of the code they commit includes AI assistance. They expect that percentage will increase considerably within the next few years. Already, 72% of those who have tried AI use it daily.

The uptick in vibe coding and AI-assisted coding amplifies the threat surface. As more developers integrate AI tools into their workflows without implementing proper verification processes, the attack surface for slopsquatting continues to expand.

For those using AI to assist with coding, double-checking output is essential. Verifying that recommended packages actually exist in official repositories before incorporating them into projects reduces risk.

Navigating AI-assisted development

Implementing automated checks that validate package names against known registries can help catch hallucinated packages before they enter production code. Security teams should also monitor for unusual package installations and maintain up-to-date threat intelligence on known slopsquatting campaigns.

Zac Amos is the Features Editor at ReHack.

Welcome to the VentureBeat community!

Our guest posting program is where technical experts share insights and provide neutral, non-vested deep dives on AI, data infrastructure, cybersecurity and other cutting-edge technologies shaping the future of enterprise.

Read more from our guest post program — and check out our guidelines if you’re interested in contributing an article of your own!

Know where to place your router and how to position its antennas should get you better coverage.

How do Wi-Fi antenna signals actually work?

Pexels: Jakub Zerdzicki

How should you position router antennas for different home layouts?

Where should you place your router for the best coverage?

Trumzz/Getty Images

A woman is bitten after her iPad is stolen, dozens of iPads are stolen from an elementary school, and a man is arrested for selling fake AirPods to police, all in this week’s Apple Crime Blotter.

The latest in an occasional AppleInsider series examining the world of Apple-related crime.

Stolen iPad leads to the arrest of a man with a federal bank robbery warrant

An iPad stolen from a victim’s car was tracked by police, leading to the arrest of a man with several outstanding warrants, including one for federal bank robbery.

According to Woodlands Online, it started when a victim in Harris County, Tex., reported the theft of items from their car. A stolen iPad was later traced to a specific address in Montgomery County.

Deputies noticed a vehicle at the residence, which Flock cameras had spotted near the scene of a Harris County burglary.

A man found at the house, identified as Michael Austin, had multiple outstanding warrants. This included a federal bank robbery warrant out of Illinois, a felony pardon and parole warrant, and a felony family violence warrant from Harris County,” the report said.

He was arrested without incident.

Man bitten by iPhone thief

A woman in Everton, in the U.K., challenged a thief who had stolen an iPad and bank card from her home, and the assailant bit her for her trouble.

According to Liverpool Echo, while the woman got the items back, “the victim was threatened and suffered a bite to her hand, causing a small gash and bleeding.”

‘Woman bitten’ after iPad stolen from Liverpool homehttps://t.co/aQTFiQUfSJ

— Liverpool Echo (@LivEchonews) July 7, 2026

The man in question did appear on CCTV video.

Two dozen iPads taken from Long Island school

Police responded in mid-June to a call that two dozen iPads had been stolen from an elementary school on Long Island.

News 12 Long Island reports the burglar entered Chatterton Elementary in Nassau County through an unlocked rear window and took 26 iPads before fleeing. In addition to the theft, police said, there was $300 in damage to window screens.

Man arrested for offering to sell fake AirPods to police

In another Long Island crime, also reported by News 12 Long Island, a man was arrested for trying to sell counterfeit AirPods to police.

It started when the 23-year-old was pulled over for a minor equipment violation when police noticed he had “several packaged Apple products” in his car. The driver claimed he sold the products as a side business and offered to sell the officers a set of AirPods for $50.

The officers were suspicious due to multiple “dead giveaways,” including that every package had the same serial number and that there were packaging errors. The man was arrested and charged with selling or attempting to sell counterfeit items.

These Long Island Apple crimes follow the high-profile hijacking of a truck carrying Apple products back in January near the Manhasset Apple Store, also on Long Island.

A man had his iPhone stolen from a “safe pickup” site

A New Jersey man ordered a new iPhone and had it delivered to a CVS store, described as a “safe pickup site.” But just because it wasn’t delivered to his porch doesn’t mean a porch pirate couldn’t get to it.

ABC 7 in New York explains a security camera recorded a thief arriving at the store, showing an ID, signing, and walking away with the iPhone that wasn’t his. The rightful purchaser of the iPhone received one message that the iPhone had been delivered and another that it had been picked up.

After 7 On Your Side looked into the matter, T-Mobile is working with the rightful owner to “resolve” the matter.

Police told the TV station they are “still looking for the suspect in the video and say he’s part of a crime ring.”

Man arrested for stealing from nursing home, trying to use stolen credit card at Apple Store

In June, a North Carolina man was arrested for entering a nursing home in Georgia, falsely claiming to work there, and stealing credit and debit cards from rooms there.

According to WSB-TV, the man later went to the Apple Store at Lenox Mall and attempted to buy an iPhone with one of the stolen credit cards, but one of the victims received an alert.

The man faces burglary charges, and also charges of exploiting the elderly, financial transaction card theft, and financial identity fraud.

Nebraska deputy’s firing involved child sex abuse material on iCloud

A county deputy in Nebraska was fired in May, and a report in July revealed why: His iCloud account was linked to “images of potential child sexual abuse material, also referred to as CSAM.”

WOWT reports the National Center for Missing and Exploited Children had received a tip the previous December. This triggered an internal investigation that resulted in his firing.

The investigation also uncovered other allegations, including that the deputy was “taking photographs of on-duty female DCSO deputies and manipulating the photos using AI to generate nude images,” as well as various allegations of intoxication and violent threats.

The deputy appealed the finding but had the appeal denied.

Indian shopkeeper, in viral video, accuses man of stealing iPhone case

At Delhi’s Connaught Place, a video went viral in July of a shopkeeper accusing a foreign tourist of taking an iPhone case from his store.

This white tourist allegedly took an iPhone case without paying and ran away in CP .

When the shopkeeper tried to stop him , he reportedly started abusing and arguing with the shopkeeper.

If thier country’s per capita income is higher , why do they behave like a beggar ? pic.twitter.com/FUo66pXlP8

— Pankaj (@Ragepkj) June 28, 2026

Advertisement

According to NDTV, the shopkeeper claims “the tourist picked up an iPhone case and tried to leave the store without making a payment. When he attempted to stop him, the tourist allegedly denied stealing anything and refused to stop.”

AI + ML

The RAMpocalypse may be the precursor to the AIpocalypse

It’s a good time to be in the memory business. As the AI datacenter business booms, SK Hynix and Micron’s revenues have tripled in the last year, and Samsung’s has roughly doubled.

But while the trio have the AI revolution to thank for their good fortune, the deck is stacked for a reversal. Such is the memory business historically.

Today, sky high demand for high-bandwidth memory (HBM), DDR5, and NAND flash memory needed for GPU servers has devoured any remaining capacity, leading to shortages that have driven up prices on everything from consumer electronics to AI infrastructure. You can’t even buy a budget smartphone these days.

The big three memory vendors are now in the process of investing hundreds of billions of dollars to bring new fab capacity online.

In June, South Korean President Lee Jae Myung announced a $576 billion investment led by SK Hynix and Samsung to bolster chip production and shore up AI supply chains.

On Thursday, Micron said that it would invest up to $3 billion to strengthen the US semiconductor supply chain, and according to recent reports, the Idaho-based chipmaker is also working to boost production across its Singapore, Taiwan, and Japan sites.

Unfortunately, it’s a slow process.

Semiconductor manufacturing is among the most complex and resource-intensive industries in the world, and building a new DRAM or NAND flash wafer fab is not a trivial endeavor.

Before the first chip can roll off the production line, financing must be secured, a location must been selected, permits must be won, and tens of millions of dollars of support facilities ranging from power conditioning and air handling to the ultra-pure water filtration systems must be deployed.

Even after the clean rooms are completed, hundreds of millions of dollars of specialized lithography, wafer transport, and test equipment must be installed and validated. And once everything is ready to be powered on, it can take months to dial everything in and bring yields to acceptable levels. This process often takes years even without delays.

So while there are a handful of new memory fabs already under way, anything SK, Samsung, or Micron starts today will take at least three years to bring online, and even longer to ramp production.

That means memory prices are going to stay high for the foreseeable future. A recent IDC report warns that we may not see relief from the RAMpocalypse until at least 2028.

That’s great news for memory makers, whose revenues will stay inflated. But it’s a big problem for AI startups and model devs, who will be paying higher infrastructure prices until that happens.

OpenAI and others have spent the last four years or so and hundreds of billions of VC capital developing ever more capable models, agents, and tools. It’s no longer a matter of whether the technology works, but rather whether the benefits justify continued investment at current or higher levels. 

Sooner or later, these startups will have to turn a profit, and sky high memory prices certainly aren’t helping to find anything resembling a margin in the cost per token.

The question now is whether or not the memory vendors can bring new capacity online before the great AI houses exhaust their VC-subsidized runway and the music stops.

Historically, memory is a commodity, with wild swings in pricing characterized by boom and bust cycles. Memory vendors therefore rely on boom cycles to finance fabs, knowing full well that, once they come online, the additional capacity could end up cratering prices.

As we reported late last year, the AI boom has changed this dynamic dramatically. Where we should have expected memory prices to fall across 2025 and 2026, we’ve seen the exact opposite as AI infrastructure consumes every bit of DRAM and NAND it can get its hands on.

But if the anticipated demand for AI falls short, everyone loses and memory vendors will find themselves at the bottom of a bust cycle to end all bust cycles.

On a bright note, the sky-high price of memory will no longer factor into why you can’t afford a new laptop or smartphone. ®

In the past few years, multiple tech executives have told us that glasses could be the next big interface for consumer hardware. And yet, today’s smart glasses rely a lot on phones, even if they have good hardware. Even Realities’ G2 smart glasses are in the same boat. They’re a premium-looking pair of glasses with a neon-style heads-up display you can see in any lighting — but their functionality relies heavily on their connectivity with the phone, which can be** unreliable and frustrating.

Even Realities takes a different approach to smart glasses than players like Meta. Their devices have a monochrome heads-up display that shows text and information in green, giving it the look of a neon board.

There are no cameras or speakers, and that is by design. The company wants to focus on productivity rather than recording, so the people around you don’t have to worry about being filmed.

The G2 is the second pair of smart glasses from Even Realities and an improvement over the G1 released a few years ago. The G2 has a brighter 1,200-nit display (vs. 1,000 nits on the G1), four mics (vs. two), and a 75% larger display area than its predecessor. The new display also has a better 60Hz refresh rate, compared with 20Hz on the G1.

In the few months I’ve used the G2, the connectivity with the phone has improved tremendously. Early on, the glasses would disconnect from the app so frequently that I nearly gave up on them. But after a few app updates, that issue got better.

The glasses are targeted at people who might be constantly in meetings, giving presentations, and traveling to countries where different languages are spoken. 

Design

The glasses, which come in two frame designs, are very light at 35 grams. The frame is made out of magnesium alloy, and the temples (the arms that go over your ears) are made out of titanium alloy. In terms of weight and fit, the glasses were comfortable to wear. 

Since I work from home most of the time, though, I didn’t feel much need to wear them all day. That said, the lenses have UV protection built in, so they’re still worth wearing outside just for eye protection — smart features or not.

The company claims that, based on typical usage, G2’s battery can last up to two days on a single charge. The glasses come with a protective case that can recharge them up to seven times before needing to be plugged in itself. I personally didn’t test the two-day claim, but the battery lasted me long enough to put them back into the case without running out of juice.

That case is big — you can’t shove it in a pocket — but it’s solid, and the glasses fit in snugly.

Features and operation

The glasses act as your companion for schedules, reminders, and access to notes. You can wake them up by tapping on the stem-based controls. If you double-tap on the control pad on the stem, you will see a dashboard with information like your upcoming meetings, stocks, and top news. 

The G2 can also show real-time phone notifications, but the pop-ups weren’t always reliable — and since my phone is usually within reach anyway, I didn’t find much use for the feature.

Long-pressing the temple control opens a menu with several functions: a notifications tray, Translate, Conversate, Teleprompt, a to-do list, and Navigate. Translate lets you set a target language and converse with anyone. At the recent Global Connect Show (GCS) in China, I wore the glasses while talking to company reps doing demos, and the translation was good enough for me to follow along when someone spoke Chinese. I also tried it with other journalists speaking various languages, including French and Spanish. (The downside of this feature is that the other person doesn’t know what you’re saying in your language unless they’re also using the app.)

Navigate is a cool feature that shows turn-by-turn directions on the heads-up display. The catch: it doesn’t work with Google or Apple Maps. Instead, you have to set your route through the Even Realities app. I tried it a few times walking to cafes near my house. The directions showed up well on the display, but the app kept getting the addresses wrong, so I can’t rely on it for places I don’t already know how to get to. Still, I could see cyclists or motorbike riders finding it useful once the company fixes the accuracy issues.

Conversate, at first, just showed a live transcript of the conversation on the glasses, which felt pointless since you can just as easily record a meeting with an app or an external notetaker. Later, the company added a “prep notes” feature that surfaces more context: you can manually add notes or documents ahead of a meeting and let the AI reference them during the conversation, or let it listen in real time and pop up short explainer bubbles for concepts as they come up. For instance, during a briefing about energy, it showed me a bubble for “Green Hydrogen,” and tapping it brought up a definition right in front of my eyes. That was genuinely useful — though I wouldn’t want a transcript or explainer bubbles for every conversation I have.

At the center of all this is the built-in assistant, Even AI. As with any voice assistant, you say a wake word to activate it and ask questions or add items to your to-do list. It often misunderstood my to-do list requests, and for general questions, the answers were often long paragraphs that streamed across the screen with no way to interrupt or skip ahead.

Another issue: despite having four mics, Even AI often failed to activate, or misheard me, when I was outside. The ambient noise in India could have played a part, but I’d still expect a modern gadget to have better noise handling.

The G2’s screen was legible in most conditions, but in a bright room I had to adjust the brightness manually through the app. Even if the company hasn’t built an automatic-brightness sensor yet, I’d like to see a manual brightness control built into the glasses themselves, rather than requiring the phone app.

Don’t put the R1 ring on it

Even launched a companion ring called the R1 alongside the G2. The idea is to control the glasses through a touch surface on the ring instead of the glasses’ own touch controls. But its price and functionality don’t quite justify the cost.

The ring works well, and I didn’t have any issues using it. But I struggled to find scenarios where I actually needed it, since the touch-sensitive temples on the glasses already do the same job.

On top of that, Even built health tracking into the ring — heart rate, calories, steps, sleep, and SpO2 (blood oxygen level). Personally, I’d rather go for a dedicated ring like Oura or Ultrahuman if I wanted that form factor with health tracking. Second, if I already use a fitness tracker, I wouldn’t want to buy a ring where health is an auxiliary function for a ring that is meant to control the glass. 

All this functionality bumps up the ring’s price to $249, which is not cheap. If I used my smart glasses a lot, I would consider buying a controller ring at a lower price if it also had a mic, which I could use for issuing commands to the AI assistant. As it stands, I’d skip the R1. 

Where does Even G2 stand?

Smart glasses are coming out fast. Camera-equipped, screen-free models like the Meta Ray-Bans are popular, but Meta, Snap, and other competitors are racing to build glasses with color screens, too. Only a handful of Chinese companies — like Rokid and Inmo — are making glasses with this same neon-display style.

The Even G2 costs $599 and delivers solid hardware in a light, good-looking frame. The company is also working to make the glasses more customizable by supporting third-party apps, though I didn’t find any app compelling enough to make me reach for the glasses more often. They’re a nice-to-have: fun to explore if you like tinkering with new hardware and don’t mind trying out third-party apps.

The hardware itself is good, but outside of jobs that require constant translation or teleprompting, it’s hard to find a clear everyday use case for smart glasses like these.

Even’s bet is that skipping the camera and speakers is the right move for a productivity-focused device — and I don’t disagree with that direction. But now that the company has newly reached unicorn status, it needs to build out more first-party software to make the glasses something people actually reach for every day.