My old 21-inch gas mower was a Toro that a friend of mine found for free at the side of the road. He gave it to me, I replaced the carburetor, and it fired right up. Score! It ran fine for over a year. And mowing 5 acres with a small gas mower at an hourly rate made my son a relatively wealthy 11-year-old. But then it died again. And again. And at some point, small engine repair isn’t fun anymore.

While I was spending weekends cleaning carbs, tracking down air filters, and not mowing, the grass was growing. In northern Wisconsin, grass grows so fast I’m pretty sure you can see it happening if you take the time. A wet June ensured that by the time the Ego arrived for testing, the grass was well out of control, up to 10 inches in the lower, wetter areas of the property.

I figured it would make a good torture test for the Ego, but I’d still need to go in afterward to hack it back with my string trimmer. To my surprise, the Ego 1300 plowed right through almost all of it in a single pass.

For the first pass, I set the height to 4.5 inches, which is easy to do using the height adjustment lever. Unlike an alarming number of mower makers, Ego’s height adjustment system is a thing of beauty. The height markings aren’t arbitrary letters or symbols that require you to consult a manual to translate to inches. On the Ego, the height adjustment is plainly printed in inches. Imagine that. This lets you know that if you set it to 4.5 inches, your grass will be 4.5 inches high when you’re done.

Photograph: Scott Gilbertson

Starting at 4.5 inches and running in Turbo mode allowed the Ego, in a single pass, to cut just about anything I pointed it toward. There is one patch of grass, below an active seeping spring, that was long enough and wet enough that I had to make two passes to get a nice even cut. After that initial mow, I then went back over it a few days later at 3 inches to finally get the height back to what I could call normal. The second mow was, dare I say, incredibly easy. Even the lowest self-propelled speed was generally fast enough for me, though I did crank it up to about the middle to get up the steeper slopes (and then turned it off entirely going downhill).

Apple is being forced to detail fees it wants to charge for outbound links in its continuing legal fight with Epic, and now it’s only got 24 hours to file them.

The ever-ongoing legal battle between Apple and Epic Games over what Apple should charge apps within the App Store has hit a snag for the iPhone maker. In the current phase, the two are fighting in the U.S. District Court for the Northern District of California over new fee proposal.

According to X posts by Epic Games founder and CEO Tim Sweeney on Tuesday, Apple attempted to delay a filing of proposed fees to the court. To Sweeney’s glee, that stay was denied.

As a consequence, Apple’s legal team now has just 24 hours to file the proposal with the court. After that Epic will have 60 days to analyze the proposal and make its own filing with the court.

Sweeney continued with a second post to X about the legal activity, asking if Apple will “honestly document their costs for human reviewer time and seek to recoup them?” Alternately, he proposes Apple may “fabricate outlandish new notions of cost previously unknown to mankind.”

As it stands, Epic and commentators have another 24-hour wait before finding that out.

How Apple got here

Apple and Epic have been in a lengthy battle that started with “Fortnite” letting players make in-app purchases using a third-party payment processor, against App Store rules. Epic also made demands that included allowing alternate app storefronts in iOS, and a change in the commission structure.

Apple did come out of the lawsuit pretty well, succeeding in many areas but failing to fend off changes to anti-steering measures, namely preventing developers from sending users to other processors to pay. Apple was ordered to make changes.

Epic convinced a court in April 2025 that Apple didn’t follow the spirit of the law, which led to more legal activity.

The latest action is midway through a schedule to discuss changes in App Store fees for outbound links, which goes back to May 2026.

Apple was given 45 days to file a “proffer,” or a good-faith offer of evidence and testimony to the court, with a 10-day timer to hand Epic non-privileged documents about the decision-making process after that. Five days later, Apple was to meet with Epic to discuss the privilege log and decide what documents warrant further review by a third party.

Today’s stay denial is at this point in the process.

Afterwards, Epic has 60 days to file its own response to the court, in a 30-page document. Apple would then have another 30 days to file a reply.

A Fortnite foible

While Sweeney’s tweets are chiefly about the lawsuit, there was an odd element about “Fortnite” and the Mac. In response to a question about what the decision means for the game on macOS, Sweeney said “We are rapidly approaching an endgame that will determine that one way or another.”

Currently, “Fortnite” is unavailable on macOS, but it seems to be less about being blocked and more Epic refusing to use the platform for its premier game.

At the time of the initial legal fight, Epic’s developer accounts were frozen out by Apple. However, in March 2024, Epic confirmed that Apple would reinstate its account in the EU.

Though celebratory in tone for the prospect of getting the game out on iOS, Epic didn’t discuss macOS at all.

As it stands, there doesn’t seem to be anything standing in Epic’s way from releasing “Fortnite” on macOS at all.

It has its own Epic Games Store on macOS, which can be used to make purchases completely outside of Apple’s App Store system. Also, developers can send Mac apps for notarization and still be distributed outside of the App Store itself, which Epic surely could handle with its reinstated developer account.

Parents who are stressed out are more likely to rely on screens and devices to keep their kids busy or calm — and less likely to enforce screen time limits. That’s the takeaway from a new screen time study from Florida International University’s Center for Children and Families. Researchers said the effects were more than a child acting out. 

Screen time is one way to keep kids busy, especially when school’s out. The time spent on tablets and laptops can be a break for parents to complete tasks or take a break. But readjusting to early mornings, class time and homework can be a big shift for both parents and kids when school is back in session. 

Jeff Temple, a professor and psychologist with UTHealth Houston, who was not affiliated with the FIU study, agrees. “For many families, screens become one of many tools that help everyone get through the day – and that’s mostly OK,” he told CNET. “The challenge comes when the structure of summer doesn’t transition back to the structure of the school year.” 

Screen time isn’t a parenting failure, but experts have concerns and recommendations to help families strike a healthy balance for the back-to-school season. 

Screen time is determined by parents’ stress

FIU’s study evaluated 822 caregivers of children ages 4 to 8 through online surveys on the child’s behaviors, parenting stress and screen time strategies. The findings show that parents opt for more screen time when a child’s behavior seems difficult and how much time is usually less enforced. 

Shayl Griffith, an assistant professor in FIU’s Department of Counseling, Recreation and School Psychology and senior author of the study, said in a news release that parents shouldn’t be afraid to ask for help. The study’s recommendation is to focus on screen time and media that are “meaningful, engaging and appropriate for their child’s age.” 

“When caregivers have the resources they need, they are better able to set consistent boundaries and engage with their children in ways that promote healthy development,” Griffith said.

The fine line between digital media discipline and disruption

Youth screen time is a growing concern. In January, the American Academy of Pediatrics published findings on correlations between digital media use and age-related development to help determine screen time rules for your child and monitor behavior.  

Toddlers and kindergarteners who use digital media for high-quality educational content can benefit from good social behaviors and language — especially when a caregiver is viewing the screen and interacting with the child. However, young children who use tablets for hours, but the AAP says too much screen time is linked to angry outbursts.

School-aged children, ages 6 to 12, can benefit from thoughtful educational digital media in moderation, the AAP study found. But again, there are risks. Too much media use can lead to lower academic achievement, attention control and cognition. Some media can be problematic for kids, like apps encouraging purchases, those focused on gaming with not much educational value and content that is dangerous or risky. And there’s a big callout included in the list: school devices that give access to “distracting video games, videos and other platforms.”

Managing stress, school and screen time is tricky

With summer wrapping up and school approaching, the big question is: How should stressed parents think about screen time? School may be the solution. 

“School can help provide a reprieve for families who do not have access to childcare in the summertime months and serve as a helpful jumping-off point to re-evaluate digital media habits, given the change in routine already,” said Dr. Tiffany Munzer, a professor of pediatrics at the University of Michigan Medical School. It can also mean more time for in-person interactions, which is important for social connection, Munzer added. 

However, making the switch from summer to back-to-school season can be challenging. And adjusting screen time while managing life stresses may not be easy. Munzer recommends making incremental changes. For example, if your child loves watching do-it-yourself videos, find a way to recreate what they’re seeing, Munzer said. Or if digital media is a way to calm kids, look for alternative activities such as reading or walking, Munzer added.

Jason Nagata, an associate professor of pediatrics at the University of California, San Francisco, said the back-to-school season is a good time to revisit family screen time rules and routines. 

“Rather than simply trying to return to an old set of rules, families can consider what makes sense for the child’s current age, developmental stage, school schedule and use of devices for schoolwork and communication,” said Nagata. “A family media plan should be dynamic; it may reasonably look different during the school year than during summer or holidays, and on weekdays compared with weekends.”

The AAP recommends creating a Family Media Plan based on your family’s needs. The plan is an agreement that lets you prioritize healthy, safe and consistent screen time boundaries. That includes creating a list of approved media kids can choose from, a schedule of screen-free time, how to choose good media content and outlets for your family. 

Experts also recommend the AAP’s 5 C’s of Media Use, tailored to your child’s age. The five C’s are Child, Content, Calm, Crowing Out and Communication. They can help you set and communicate screen time guidelines based on your child, their emotions, the content and how it affects them. That may not help the stress that parents face, but it could help regulate your child’s screen time without guilt.

Nagata encourages parents to pay attention to their children’s time and relationship with screen time. Look out for warning signs like difficulty cutting back on screen time despite trying, becoming preoccupied with social media or a phone, experiencing conflict because of screen use and using screens to escape problems. And pay attention to see if screen use is interfering with sleep, school relationships or other activities so you can make more adjustments sooner. 

Dashia is the consumer insights editor for CNET. She specializes in data-driven analysis and news at the intersection of tech, personal finance and consumer sentiment. Dashia investigates economic shifts and everyday challenges to help readers make well-informed decisions, and she covers a range of topics, including technology, security, energy and money. Dashia graduated from the University of South Carolina with a bachelor’s degree in journalism. She loves baking, teaching spinning and spending time with her family.

See full bio

Advertisement

Prop Department set a simple goal for a recent build: take a full-size Tesla Cybertruck and turn it into something anyone could drive from a distance. There is no driver in the seat, and there are no complex software overrides of the truck’s own systems. Instead, there are motors, linkages, and painstaking wiring that allow someone outside the vehicle to handle the steering, throttle, transmission, and brakes. The project belonged to WhistlinDiesel, whose videos often end with expensive machinery in pieces, so the finished truck needed to work well enough for a high-speed run that might not go gently.

Three days passed, and that was all the calendar allowed, as the parts were fashioned from what was already in the shop rather than ordering new ones. To make more space, the front seats were pushed out. A special frame with laser-cut brackets was installed next, all screwed down tightly to ensure it stayed there even when things got bumpy. A stepper motor was mounted to the steering wheel using a previously constructed adapter flange that matched the Cybertruck’s column. That motor handled the left and right turns naturally. A couple of servo motors took care of the rest, with one kicking the accelerator pedal, another working the small fingers on the gear selector buttons, and a ground wire dealing with the capacitive sensors. Then there were the power supply, which used four different voltages to keep everything running. The stepper motor received 48 volts from the truck’s mid-voltage battery. Smaller transformers reduced the voltage to 24 volts for the servos, 12 volts for the relay bank, and 5 volts for the Arduino that controlled the display.

Sale

DJI Neo, Mini Drone with 4K UHD Camera for Adults, 135g Self Flying Drone that Follows You, Palm Takeoff…

• Due to platform compatibility issue, the DJI Fly app has been removed from Google Play. DJI Neo must be activated in the DJI Fly App, to ensure a…
• Lightweight and Regulation Friendly – At just 135g, this drone with camera for adults 4K may be even lighter than your phone and does not require FAA…
• Palm Takeoff & Landing, Go Controller-Free [1] – Neo takes off from your hand with just a push of a button. The safe and easy operation of this drone…

The brakes were a challenge to fix since a pneumatic cylinder kept the foot pushed down at all times, waiting for a remote signal to let it go again.Because the OEM air tank couldn’t stay charged for long, an air compressor had to ride shotgun in the backseat. Without the ‘dead-man’ setup, the truck might just continue rolling away when our signal failed. One final stumbling block remained: the cabin sensors, which were designed to identify a live human in the driver’s seat, would shift the truck into Park as soon as that seat was empty. The solution was to simply tear off the module, and the truck stopped insisting on a body in the wheelchair.

Now that the FPV camera was mounted on a strong bracket, the driver could see clearly via their goggles. In early tests, the remote signals got mixed up with a racing simulator’s motion platform and gyroscope, but screen slowness and HDMI peculiarities put the notion on hold for a later edition. Uncomplicated radio control worked flawlessly, with steering responding, throttle kicking in, gears shifting, and the emergency brake locking up just as instructed.

WhistlinDiesel drove the finished truck to a private race track. Full acceleration down the strip pushed the speedometer to 111 mph when the Cybertruck smashed into a motionless Ford F-150 parked broadside. The Ford’s cab was ripped completely off its frame in a firestorm. The Cybertruck crumpled violently in the front, yet remained intact and did not even light its battery pack. Both trucks were scrapped at the end of the day, but the remote system did exactly what it was supposed to do.

Things have already taken a near-fatal turn in Lioness season 3, which is something I didn’t expect to happen so soon. Indeed, last week, we saw two unnamed Russian agents posing as police officers turn up at Joe’s door and demand to enter her home, only for them to be killed in the process.

The entire team is called out to investigate, but when a group of “watchers” are spotted on a nearby roof, the Lionesses wonder if it’s a trap… if they’re here, they’re not looking elsewhere.

When I first downloaded the McDonald’s app years ago, I signed up for its loyalty program hoping to get better deals. Why not get some cheap French fries? While I understood that this would entail some type of data tracking, I didn’t grasp how the information would be used to algorithmically predict my next purchase.

As a California resident, I have the legal right to access the data a company stores about me. So, I decided to request that information from McDonald’s to better understand what one of the major fast-food companies is collecting about its customers. A few days after I visited the McDonald’s Privacy Rights Center site and requested access, I received a 515-page file in my inbox with the iconic golden arches stamped across the top.

When customers sign up for loyalty programs, like the one offered through the McDonald’s app, they might not fully realize the extent to which their data is being aggregated and used in predictive models.

“McDonald’s secret sauce is really commercial surveillance,” says Jeff Chester, executive director at the Center for Digital Democracy, a group that advocates for consumer protections. Privacy experts I spoke with said this level of detail may feel invasive, but it’s fairly standard for how large companies in America run their loyalty programs.

“This report contains specific pieces of personal information about you that were identified by searching McDonald’s systems which contain information about our customers,” read the report’s intro. It was lengthy and difficult to parse, so I used a generative AI tool to extract key information before verifying details with the original document and reaching out to privacy experts.

“McDonald’s takes data privacy and security seriously, and we take robust steps to safeguard customer information,” a McDonald’s spokesperson tells WIRED via email. “Like many digital loyalty programs, we use information such as past purchases to provide a more engaging, personal customer experience—like delivering the most relevant deals, offers and messages. Our customers continue to have privacy choices available to them as outlined in our privacy statement.”

Much of my data report contains a detailed account of past McDonald’s transactions as well as offers the fast food company sent me and how many loyalty points I accumulated. Essentially, every time I opened up the app to grab a bite to eat, it created a detailed record about when, where, and what I purchased. The log was even more thorough than I anticipated, including a record of every time I’ve scanned a code as part of its returning Monopoly sweepstakes along with the prize I received.

Ryanair has signed a five-year data and AI partnership with Google Cloud, rolling out Google Workspace and Google Cloud services to 35,000 employees and making Gemini Enterprise, Google’s agentic AI platform, the centrepiece of how Europe’s largest carrier intends to run itself.

The airline has tied the deal explicitly to its growth target of 300 million passengers a year by 2034, roughly a 44% increase on the 208.4 million it carried in the financial year to March.

Gemini Enterprise will be used, in the companies’ description, to automate decision-making, optimise flight crew logistics, and support corporate productivity generally.

Ryanair plans to use Google DeepMind models, naming AlphaEvolve and WeatherNext, to support fleet operations and maintenance scheduling.

WeatherNext is DeepMind’s forecasting family, which for an airline running around 3,900 flights a day out of 95 bases is not a peripheral application. It also plans to replace its existing collaboration systems outright with Workspace and Gemini.

Running underneath all of it is a resilience argument that Ryanair has put ahead of the AI one. The airline describes the arrangement as a “dual-cloud strategy”, in which Google Cloud helps build a system flexible enough that if one provider has problems, critical services keep running.

“Ryanair is on an incredible growth journey to 300 million passengers by 2034,” said Eddie Wilson, Ryanair’s chief executive.

“To support this growth, we need to ensure we have excellent infrastructure resilience, and our new dual-cloud strategy provides this, alongside technology partners that match our speed and relentless focus on efficiency.”

Maureen Costello, Google Cloud’s vice-president for the UK, Ireland, and sub-Saharan Africa, put the case in industry terms. “Aviation is an industry defined by precision, and Ryanair is a pioneer in operational execution,” she said.

“This agreement demonstrates how deploying generative AI at scale, coupled with modern collaboration tools for frontline workers, can help industry leaders scale securely, reduce operational costs, and redefine the travel experience.”

The scale Ryanair is buying for is real. It ended the last financial year with 647 aircraft, having taken delivery of all 210 Boeing 737-8200 Gamechangers, and has 300 737 MAX-10s on order, split evenly between firm orders and options.

Boeing expects certification of that aircraft late this summer, with the first fifteen due to Ryanair in spring 2027. Pre-exceptional profit after tax for the year was €2.26bn, up 40%.

What the release does not say is as notable as what it does. There is no named migration of specific workloads, no BigQuery or Vertex AI commitment, no disclosed contract value, and no timeline for when any of the AI applications reach production.

Airlines have been announcing operational AI for the better part of a decade; the difference here is the size of the seat count and the explicit link to a passenger target eight years out.

Ryanair is not new to building software, whatever its reputation for spending nothing.

It runs Ryanair Labs, an in-house technology operation with hubs in Dublin, Madrid, and Wrocław, and has spent years pushing passengers towards direct booking, a campaign that produced a €255m Italian antitrust fine in December 2025 over its treatment of online travel agents and a settlement with Booking Holdings the previous August.

It also runs against a current. Airbus is moving its most critical applications off AWS to a French sovereign cloud, a decision framed around European control of European infrastructure. Ryanair, characteristically, has gone the other way and bought the American stack twice over.

A hot potato: A proposed Amazon data center in Texas has sparked concerns that its massive on-site power plant could become the single-largest source of pollution in the US, threatening to affect local communities, farmland, and small businesses. The facility is planned for Pecos County as part of Amazon’s growing push into AI and cloud services.

In a statement to The New York Times, an Amazon spokesperson said the data center will be powered entirely by an on-site power plant, ensuring it won’t raise electricity costs for local residents. The plant is expected to run on natural gas, with 35 turbines combining to generate up to 7.65 gigawatts of power.

The facility has regulatory approval to release up to 33 million tons of carbon dioxide per year, which would make it the largest single source of greenhouse gas emissions in the US – more than any other factory or power plant nationwide.

(Not Amazon’s data center) – Satellite imagery shows OpenAI’s Stargate data center transforming the landscape near Abilene, TX. The 1,100-acre campus spans roughly 4 million square feet and is still growing. Credit: Planet Labs PBC

Critics say the permitted emissions level is far too high, noting that the projected output would rival the total emissions of entire countries, including Switzerland, Ireland, and Bulgaria.

Notably, Amazon co-founded The Climate Pledge in 2019, promising to reach net-zero carbon emissions by 2040. Since then, though, its emissions have climbed sharply: the company’s emissions tied to purchased electricity rose 34% in 2025 alone.

Stargate’s gas turbines are permitted to emit over 1.6 million tons of greenhouse gases and 1,000 tons of air pollutants annually, approved without public input. Credit: Evan Simon / Floodlight, via The Texas Tribune

Asked about the pledge, Amazon spokeswoman Margaret Callahan told the Times that “the world looks different now than when we co-founded the climate pledge,” but added that “our commitment [to a carbon-neutral future] hasn’t changed.”

The Pecos County plant falls under the so-called “Ratepayer Protection Pledge,” which requires tech companies to source their own power for new AI data centers rather than drawing from existing grids.

The pledge was signed last March at the insistence of President Trump by several leading American tech companies, including Google, Microsoft, Meta, Amazon, Oracle, OpenAI, and xAI.

US investment in data center infrastructure hit record levels last year, and as major tech companies continue pouring money into AI, demand for electricity is expected to keep climbing sharply in the years ahead.

Across 107 enterprises, agentic orchestration is not a choice of a single platform.

The typical enterprise runs three orchestration platforms at once, and selects them for flexibility across models rather than affinity to any single one. Microsoft leads primary usage while Anthropic leads forward consideration by a wide margin. 

The AI control plane enterprises expect is deliberately hybrid, meaning it includes use of the leading AI providers, but also provider-independent technologies — and the risk they fear most from provider-resident control is not lock-in but the provider’s own security and permissioning limits. 

One in five enterprises still has no real-time way to stop a runaway agent before the bill arrives.

This wave of VentureBeat Pulse Research examines enterprise agent orchestration: which platforms enterprises run on, what drives the choice, what they optimize for, how they expect agent control to be structured, and — most revealingly — how orchestrated their deployed “agents” actually are and how tightly they control the cost of running them.

The central finding is that orchestration has become plural. Eighty-five percent of enterprises run two or more orchestration platforms and 64% run three or more, with a mean of 3.1 platforms per organization. Microsoft AI Foundry / Copilot Studio appears in 70% of stacks and OpenAI’s Agents SDK in 68%, with Anthropic’s Claude Platform in 47%. Asked to name a single primary platform, respondents who gave one unambiguous answer put Microsoft first (41%) and Anthropic second (28%). Nobody in this sample is running one orchestration layer and calling it a strategy.

The selection logic follows from that plurality. Flexibility across models and tools is the leading purchase driver at 29%, nearly three times the share naming model gravity — native alignment with a state-of-the-art base model — at 10%. Enterprises are not choosing the orchestration environment that comes with their favorite model; they are choosing the one that does not commit them to any model. Security and permissions (17%), production reliability (15%), and control over agent execution (15%) fill out a buying logic focused on governance and optionality rather than developer convenience.

A clear majority (53%) expect a hybrid control plane by the end of 2026 — provider-native plus external orchestration — and the risk they most associate with provider-resident control is security and permissioning limitations (37%), ahead of vendor lock-in (23%) and limited visibility (22%). Investment has moved accordingly: agent monitoring and debugging leads the spend at 31%, with security and permissions enforcement at 30%, while workflow tooling draws 19%. Enterprises are spending to see and govern agents, not merely to build them.

Most companies admit that a majority of their “agents” are really just chatbots. A plurality of 47% of respondents say that between 26 and 50% of their agents are genuinely orchestrated, with 37% at a quarter or below and 16% past the halfway mark. 

But fiscal control remains the soft spot: 21% of enterprises track agent spend only through post-hoc logs, with no real-time way to halt a runaway execution loop.

Methodology

VentureBeat fielded this survey as part of its ongoing Pulse Research series, with this instrument focused on enterprise agent orchestration. Responses are filtered to organizations with 100 or more employees (n=107), drawn from a single July 2026 wave; because this is one wave rather than a pooled multi-month sample, the report reads cross-sectionally and does not infer month-over-month trends. All figures in this report come from the July fielding only. Where questions were multiple-select, shares can sum to more than 100%.

This wave draws a notably large-enterprise, technology-heavy sample, and that shapes every finding in it. By organization size, more than half sit at 10,000 employees or above: 50,000+ (26%) and 10,000–49,999 (25%) lead, followed by 2,500–9,999 and 500–2,499 (19% each) and 100–499 (11%). Technology/Software accounts for 53% of respondents, with Government/Public Sector (16%) and Manufacturing/Industrial (10%) next. By role the sample is hands-on and technical: software and ML engineers (22%), product and program managers (21%), directors of data/AI/analytics (17%), and VPs of data/AI/analytics (12%). On purchasing, 90% are recommenders, influencers, or final decision-makers for AI solutions (63% recommender/influencer, 27% final decision-maker).

A note on the primary-platform question. Forty-six of 107 respondents registered more than one selection on a question intended to capture a single primary platform. Because those responses cannot be resolved to one answer, primary-platform shares are reported on the 61 respondents who gave a single unambiguous answer, and are labeled as such wherever they appear. Platform footprint figures — which platforms an enterprise uses at all — use the full n=107 base and are unaffected. The ambiguity is worth noting on its own terms: on a question asking for one platform, more than four in 10 respondents could not or would not narrow to one, which is consistent with the multi-platform pattern documented in Finding 1.

At 107 respondents the sample is robust enough to read directionally with reasonable confidence, though it remains self-selected and is not a probability sample. Because each subgroup here only includes about 50 to 60 respondents, splits between them are less precise than the full-sample findings.

Finding 1: Orchestration is a portfolio, not a platform

The typical enterprise runs three orchestration platforms at once

We asked which agent orchestration platforms enterprises use, and which one they treat as primary. The first answer is that almost nobody has just one.

The defining feature of this layer is plurality. Only 15% of enterprises run fewer than two orchestration platforms; the median organization runs three, and one in six runs five or more. Read that way, the platform “shares” below describe overlapping deployments rather than a divided market — Microsoft and OpenAI each appear in roughly seven of ten stacks precisely because most stacks have room for several.

Asked to name one primary platform, the 61 respondents who gave a single unambiguous answer put Microsoft AI Foundry / Copilot Studio first at 41%, Anthropic’s Claude Platform second at 28%, LangChain / LangGraph at 10%, and OpenAI’s Agents SDK at 7%, with Google, Amazon, Salesforce, and custom in-house builds at 3% each. Microsoft’s lead on primary usage alongside OpenAI’s near-equal footprint on any usage is the signature of an enterprise-weighted sample: the Microsoft platform arrives through an existing enterprise agreement and becomes the default seat of record, while other platforms are added around it for specific work.

A note on reading these shares: As described in the methodology section, the respondents are self-selected, this wave skews heavily toward large technology organizations, and the primary-platform figures rest on a 61-respondent subset. The numbers measure where this cohort has placed its orchestration bets today, within a self-selected audience of AI-active technical practitioners. A sample built this way can diverge substantially from spend-weighted market measures, and each VB Pulse survey draws its own sample with its own company-size and industry mix, so vendor figures should not be compared across our surveys, either.

Respondents rate the platforms they run at 4.17 out of 5 for overall satisfaction, 3.91 for ease of implementation, and 3.63 for value for money — with value for money the weakest of the three by a clear margin. That ordering is itself a finding: enterprises are broadly happy with what these platforms do and distinctly less happy with what they cost, which is the same nerve the fiscal-control finding touches at the end of this report. Satisfaction sits alongside a two-thirds intent to change platforms within the year; this remains a layer enterprises work with rather than settle on.

Finding 2: Flexibility, not model gravity, drives selection

Enterprises buy the orchestration layer that doesn’t commit them

We asked what most influenced the orchestration platform choice, and optionality leads by a distance.

Flexibility across models and tools (29%) is the selection-side explanation for the multi-platform reality in Finding 1: enterprises are choosing orchestration environments on the strength of what they leave open rather than what they lock in. Model gravity — picking the orchestration layer that comes with a preferred frontier model — draws just 10%, less than a third of the flexibility share, which places the pull of any single base model well down the list of what actually decides this purchase.

The next tier reinforces the governance emphasis. Security and permissions (17%), production reliability (15%), and control over agent execution (15%) together account for 47% of responses: nearly half of enterprises pick their orchestration platform on whether they can constrain and depend on what it runs. Ease of development draws 8% and total cost of ownership 4%, an inversion of how these platforms are usually discussed in engineering circles. Performance sits last at 2% — at this stage of adoption the binding constraints are optionality and control, not raw speed.

Finding 3: The job is reliable multi-step execution

Enterprises judge orchestration by whether it completes the work

We asked what enterprises optimize for — their primary success metric for orchestration. Reliability and multi-step workflow management lead, with developer productivity closer behind than in the buying criteria.

Task completion reliability (30%) and multi-step workflow management (27%) together account for 57% of responses: orchestration succeeds, in the enterprise view, when it reliably carries a task through multiple steps to completion. Developer productivity takes a substantial 23% — notably higher than ease of development’s 8% as a purchase driver in Finding 2, which suggests enterprises do not expect to buy developer velocity so much as to earn it once the platform is in place. End-user experience is a minor concern at 7%, consistent with orchestration being an internal execution problem rather than a UX one.

This reliability-first standard is the yardstick against which the portfolio-maturity finding later in this report should be read: enterprises define success as dependable multi-step execution, and a little over a third of them still say a quarter or fewer of their deployed agents do multi-step work at all.

Finding 4: Two-thirds plan to move — and Anthropic leads the consideration set

The installed base and the pipeline point to different vendors

We asked whether enterprises plan to adopt a new, additional, or replacement orchestration platform in the next 12 months, and which platforms they are considering.

Two-thirds of enterprises (67%) intend to adopt a new, additional, or replacement orchestration platform within the year, but the clock runs longer than the intent suggests: the largest cohort sits at 6–12 months (28%) and only 15% expect to move within a quarter. This is deliberate re-platforming on a planning horizon, not urgent churn.

The consideration set is where this finding earns its headline. Among the 72 enterprises in motion, Anthropic leads at 43% — well ahead of Google (31%), custom in-house builds (31%), OpenAI (25%), LangChain / LangGraph (17%), and Microsoft (17%). Set that against Finding 1, where Microsoft leads primary usage and appears in 70% of stacks: the installed base and the forward pipeline point at different vendors. Anthropic draws roughly two and a half times Microsoft’s forward consideration despite trailing it on current primary usage, and custom in-house control planes draw as much interest as any external platform besides Anthropic. A further 18% of movers are evaluating with no shortlist at all.

Read alongside the flexibility-first selection logic in Finding 2, the shape of the next twelve months is legible: enterprises expect to add rather than replace, they are shopping for platforms that preserve model choice, and a substantial minority intend to solve the problem themselves rather than buy it.

Finding 5: Investment flows to watching and governing agents

Monitoring and permissions lead the spend; workflow tooling trails

We asked which orchestration-related investment will grow most next year. Observability and governance take the top two places.

Monitoring and debugging (31%) and security and permissions enforcement (30%) are effectively tied at the top and together account for 61% of planned growth. The money is going to seeing what agents do and constraining what they are allowed to do — the two capabilities that matter once agents are running in production rather than being built toward it. Workflow tooling (19%) and scaling infrastructure (18%) trail, and almost no one is standing still: just 3% report a flat budget.

The emphasis is consistent with the buying logic in Finding 2, where security and permissions was the second-ranked selection factor, and with the control-plane architecture in Finding 6. Enterprises that have decided to run agents across three platforms have a visibility and permissioning problem by construction, and they are funding it directly.

Finding 6: The control plane will be hybrid — and security is why

Enterprises split control, and fear the provider’s permissioning more than lock-in

We asked where enterprises expect the primary control plane for agents to live by the end of 2026, and what worries them most if that control sits inside a model-provider platform.

Hybrid control is the dominant expectation by a wide margin (53%). Taken together, the hybrid, custom in-house, and externally-abstracted options — every architecture that keeps control at least partly outside the provider — sum to 78% of enterprises, against 14% willing to hand control to a provider-managed service outright.

The reason enterprises give is worth separating from the one usually assumed. Security and permissioning limitations lead the risk question at 37%, well ahead of vendor lock-in at 23%, with limited visibility and observability close behind at 22%. Combining the security and visibility answers, 59% of enterprises name a control-and-oversight concern rather than a commercial one. The worry is less that a provider platform will be hard to leave than that it will not let them see or constrain what their agents are doing while they are on it — the same concern funding the monitoring and permissions spend in Finding 5. Only 2% say provider-resident control is not a concern at all.

Finding 7: The chatbot trap is loosening, not broken

“Bridging the gap” is now the modal answer on portfolio maturity

We asked enterprises to assess their portfolios honestly: What share of their deployed “agents” are true multi-step orchestrated workflows versus simple single-prompt chatbot wrappers.

The center of gravity has moved into the middle band. Just under half of enterprises (47%) now put between a quarter and half of their portfolio in genuinely orchestrated, stateful workflows, and 16% are past the halfway mark. The bottom two bands — a quarter or fewer genuinely orchestrated — account for 37%, and outright pure-chatbot portfolios have nearly vanished at 3%. Against the reliability-first success standard in Finding 3, this is a portfolio that has started to do the work the orchestration layer exists for, without most of it being there yet.

Maturity tracks platform count. Enterprises reporting a quarter or less genuine orchestration run 2.8 platforms on average; those in the 26–50% band run 3.5. The organizations furthest into real multi-step work are the ones running the most orchestration platforms at once, which is the practical case for the flexibility-first selection logic in Finding 2 — multi-step portfolios appear to accumulate platforms rather than converge on one.

One split that might be expected does not appear. Organization size makes no difference to portfolio maturity in this wave: 38% of enterprises at 10,000+ employees report a quarter or less genuine orchestration, against 37% of smaller ones, and the shares past the halfway mark are equally close (16% and 15%). Whatever separates the mature portfolios from the immature ones here, it is not headcount.

Finding 8: Fiscal control is still reactive for one in five

A fifth of enterprises learn about a runaway agent from the logs

Finally, we asked how enterprises enforce fiscal control over agent token consumption — the risk that an autonomous loop exhausts a budget before anyone intervenes. The approaches split four ways, fairly evenly.

One in five enterprises (21%) has no real-time, programmatic way to stop an agent before a budget-breaking bill arrives — they learn of it from the logs afterward. Another 30% lean entirely on the native caps and throttles built into their primary platform, a control only as good as the provider’s tooling and one that sits awkwardly beside the hybrid, keep-control-outside posture of Finding 6. Roughly half of enterprises — those building custom gateways (25%) or exploiting cross-model routing to arbitrage cost (24%) — are treating token burn as an engineering problem to be controlled deterministically, and the routing group is doing so in a way that only works because they run several platforms at once.

Unlike previous waves, no size split appears here: 18% of enterprises at 10,000+ employees exercise only reactive control against 23% of smaller ones, a difference well within sample noise. The gap in fiscal control in this wave is not between large and small enterprises but between those that have built a cost-control plane and those still relying on whatever their provider ships. Read against the satisfaction scores in Finding 1 — where value for money was the weakest of three ratings at 3.63 — the picture is of a cohort that is unhappy about what agents cost and, in half of cases, not yet instrumented to do much about it.

The bottom line: Plural by design, governed by intention, metered by hope

Organizations with 100 or more employees describe an orchestration strategy built around optionality rather than commitment. They run three platforms on average, choose them for flexibility across models rather than affinity to any one, and judge them on whether they carry multi-step work reliably to completion. Microsoft anchors the installed base and appears in seven of ten stacks; Anthropic leads forward consideration by a wide margin among the two-thirds planning a change; and a substantial minority intend to build their own control plane rather than buy one. Today’s footprint describes where these enterprises are, and clearly does not describe where they intend to stay.

The governance posture is deliberate and consistent. A hybrid control plane is the majority expectation, 78% intend to keep control at least partly outside the provider, and the reason is not commercial but operational — security and permissioning limits (37%) and limited visibility (22%) outrank vendor lock-in (23%) as the fear attached to provider-resident control. The budget follows the fear: monitoring and debugging and security and permissions enforcement together take 61% of planned investment growth, ahead of the tooling used to build agents in the first place.

Where the strategy thins out is cost. Portfolio maturity has moved into the middle — 47% now report between a quarter and half of their agents genuinely orchestrated, and pure-chatbot portfolios have nearly disappeared — but 21% still cannot stop a runaway agent in real time, another 30% depend on whatever caps their provider ships, and value for money is the lowest-rated attribute of the platforms they run. Enterprises have worked out how they want agents governed well before they have worked out how to meter them.

At 107 respondents in a single July wave, skewed toward large technology organizations, this reads as a clear directional signal rather than a precise measurement. The questions for subsequent waves are whether the middle band of portfolio maturity keeps climbing, whether the forward consideration for Anthropic and for in-house control planes converts into deployment, and whether fiscal control catches up to a cost that enterprises already say they are not getting their money’s worth on.

Based on survey responses from 107 qualified enterprise respondents (100+ employees), drawn from a single July 2026 wave. This is a self-selected sample rather than a probability sample, and figures should be read directionally rather than as precise measurement. Respondents include software/ML engineers, product/program managers, directors and VPs of data/AI/analytics, enterprise architects, and directors of engineering/IT, across technology/software, government/public sector, manufacturing/industrial, and financial services organizations.