TheChek News Flow

With its new serum, the medical grade skin care brand is targeting laxity, which many weight loss drug users experience.

SkinCeuticals is betting on the GLP-1 market.

On Monday, the medical-grade skin care revealed the launch of its A.G.E. Interrupter Ultra Serum, $185, which was developed for and tested on GLP-1 weight loss drug users. While the brand declined to share sales figures, the team said this launch is expected to be its biggest of the year, driving new customer acquisition.

The serum, which joins a lineup of A.G.E. Interruper products including a cream and eye cream, is formulated with 30 percent proxylane and 4.6 percent wild fruit flavonoids to target glycation, which typically leads to aging skin, and support collagen production.

With these benefits in mind, the product was specifically formulated for GLP-1 users, as rapid weight loss often leads to skin laxity. According to experts, weight loss from a GLP-1 could also lead to a lack of hydration and collagen and lead to looser skin that amplifies fine lines and wrinkles. The SkinCeuticals team, which focuses heavily on the crossover between medical aesthetics and complementary topical products, saw the growing opportunity to tap into the GLP-1 market through skin care solutions. Specifically, 30 million in the U.S. by 2030 could be on a weight loss drug, according to JP Morgan Research.

“We are laser-focused on medical trends driving consumers to clinics, and that’s how we got the idea to test this A.G.E. Interrupter Ultra Serum with GLP-1 patients,” said SkinCeuticals’ general manager Tara Pyle. “Today is about leveraging the best of science and technology, which is also about the world of aesthetic procedures, including GLP-1s alongside skin care to maximize results. We as a brand are still very driven by scientific breakthroughs. This is the heart of all of our product development.”

As the team has been eyeing the market, Pyle also emphasized that the channels to get prescribed a GLP-1 are growing, particularly as medical spas bet on the drug.

“In some outlets, it’s the number one aesthetic service, which is fascinating. It’s in our channel,” she said. “Weight loss is also a gateway to medical aesthetics involvement. We saw a recent McKinsey paper that [said] 63 percent of GLP-1 patients who start aesthetics do it for the first time after their weight loss….[According to the] American Society of Plastic Surgeons, 41 percent of patients who are prescribed GLP-1s are considering nonsurgical procedures.”

Additional research from Circana showed that GLP-1 users overindex in a variety of categories, including beauty.

“We tested the regimen with A.G.E. Interrupter Advanced Cream and the new Ultra Serum [on 25 male and female particpants], and we noticed that after [using] the regimen, before even procedures, it visibly reduced skin laxity in just four weeks,” Pyle said. “We also tested with a minimally invasive ultrasound procedure on GLP-1 patients. This also showed an additional reduction in skin laxity by 20 percent.”

As SkinCeuticals looks to future launches, Pyle said the impacts of GLP-1s on the skin will continue to be an area of interest for the brand.

“GLP-1s [are] not going anywhere. It is a major segment of our medical customers, so it will always be a consideration,” she said.

DAX:

MDAX:

SDAX:

How OpenAI’s first reasoning model sparked Silicon Valley’s obsession with general-purpose AI agents — and what comes next.

Shortly after Hunter Lightman joined OpenAI as a researcher in 2022, he watched his colleagues launch ChatGPT, one of the fastest-growing products ever. Meanwhile, Lightman quietly worked on a team teaching OpenAI’s models to solve high school math competitions.

Today that team, known as MathGen, is considered instrumental to OpenAI’s industry-leading effort to create AI reasoning models: the core technology behind AI agents that can do tasks on a computer like a human would.

“We were trying to make the models better at mathematical reasoning, which at the time they weren’t very good at,” Lightman told TechCrunch, describing MathGen’s early work.

OpenAI’s models are far from perfect today — the company’s latest AI systems still hallucinate and its agents struggle with complex tasks.

But its state-of-the-art models have improved significantly on mathematical reasoning. One of OpenAI’s models recently won a gold medal at the International Math Olympiad, a math competition for the world’s brightest high school students. OpenAI believes these reasoning capabilities will translate to other subjects, and ultimately power general-purpose agents that the company has always dreamed of building.

ChatGPT was a happy accident — a lowkey research preview turned viral consumer business — but OpenAI’s agents are the product of a years-long, deliberate effort within the company.

“Eventually, you’ll just ask the computer for what you need and it’ll do all of these tasks for you,” said OpenAI CEO Sam Altman at the company’s first developer conference in 2023. “These capabilities are often talked about in the AI field as agents. The upsides of this are going to be tremendous.”

Whether agents will meet Altman’s vision remains to be seen, but OpenAI shocked the world with the release of its first AI reasoning model, o1, in the fall of 2024. Less than a year later, the 21 foundational researchers behind that breakthrough are the most highly sought-after talent in Silicon Valley.

Mark Zuckerberg recruited five of the o1 researchers to work on Meta’s new superintelligence-focused unit, offering some compensation packages north of $100 million. One of them, Shengjia Zhao, was recently named chief scientist of Meta Superintelligence Labs.

The rise of OpenAI’s reasoning models and agents are tied to a machine learning training technique known as reinforcement learning (RL). RL provides feedback to an AI model on whether its choices were correct or not in simulated environments.

RL has been used for decades. For instance, in 2016, about a year after OpenAI was founded in 2015, an AI system created by Google DeepMind using RL, AlphaGo, gained global attention after beating a world champion in the board game, Go.

Around that time, one of OpenAI’s first employees, Andrej Karpathy, began pondering how to leverage RL to create an AI agent that could use a computer. But it would take years for OpenAI to develop the necessary models and training techniques.

By 2018, OpenAI pioneered its first large language model in the GPT series, pretrained on massive amounts of internet data and a large clusters of GPUs. GPT models excelled at text processing, eventually leading to ChatGPT, but struggled with basic math.

It took until 2023 for OpenAI to achieve a breakthrough, initially dubbed “Q*” and then “Strawberry,” by combining LLMs, RL, and a technique called test-time computation. The latter gave the models extra time and computing power to plan and work through problems, verifying its steps, before providing an answer.

This allowed OpenAI to introduce a new approach called “chain-of-thought” (CoT), which improved AI’s performance on math questions the models hadn’t seen before.

“I could see the model starting to reason,” said El Kishky. “It would notice mistakes and backtrack, it would get frustrated. It really felt like reading the thoughts of a person.”

Though individually these techniques weren’t novel, OpenAI uniquely combined them to create Strawberry, which directly led to the development of o1. OpenAI quickly identified that the planning and fact checking abilities of AI reasoning models could be useful to power AI agents.

“We had solved a problem that I had been banging my head against for a couple of years,” said Lightman. “It was one of the most exciting moments of my research career.”

With AI reasoning models, OpenAI determined it had two new axes that would allow it to improve AI models: using more computational power during the post-training of AI models, and giving AI models more time and processing power while answering a question.

“OpenAI, as a company, thinks a lot about not just the way things are, but the way things are going to scale,” said Lightman.

Shortly after the 2023 Strawberry breakthrough, OpenAI spun up an “Agents” team led by OpenAI researcher Daniel Selsam to make further progress on this new paradigm, two sources told TechCrunch. Although the team was called “Agents,” OpenAI didn’t initially differentiate between reasoning models and agents as we think of them today. The company just wanted to make AI systems capable of completing complex tasks.

Eventually, the work of Selsam’s Agents team became part of a larger project to develop the o1 reasoning model, with leaders including OpenAI co-founder Ilya Sutskever, chief research officer Mark Chen, and chief scientist Jakub Pachocki.

OpenAI would have to divert precious resources — mainly talent and GPUs — to create o1. Throughout OpenAI’s history, researchers have had to negotiate with company leaders to obtain resources; demonstrating breakthroughs was a surefire way to secure them.

“One of the core components of OpenAI is that everything in research is bottom up,” said Lightman. “When we showed the evidence [for o1], the company was like, ‘This makes sense, let’s push on it.’”

Some former employees say that the startup’s mission to develop AGI was the key factor in achieving breakthroughs around AI reasoning models. By focusing on developing the smartest-possible AI models, rather than products, OpenAI was able to prioritize o1 above other efforts. That type of large investment in ideas wasn’t always possible at competing AI labs.

The decision to try new training methods proved prescient. By late 2024, several leading AI labs started seeing diminishing returns on models created through traditional pretraining scaling. Today, much of the AI field’s momentum comes from advances in reasoning models.

In many ways, the goal of AI research is to recreate human intelligence with computers. Since the launch of o1, ChatGPT’s UX has been filled with more human-sounding features such as “thinking” and “reasoning.”

When asked whether OpenAI’s models were truly reasoning, El Kishky hedged, saying he thinks about the concept in terms of computer science.

“We’re teaching the model how to efficiently expend compute to get an answer. So if you define it that way, yes, it is reasoning,” said El Kishky.

Lightman takes the approach of focusing on the model’s results and not as much on the means or their relation to human brains.

“If the model is doing hard things, then it is doing whatever necessary approximation of reasoning it needs in order to do that,” said Lightman. “We can call it reasoning, because it looks like these reasoning traces, but it’s all just a proxy for trying to make AI tools that are really powerful and useful to a lot of people.”

OpenAI’s researchers note people may disagree with their nomenclature or definitions of reasoning — and surely, critics have emerged — but they argue it’s less important than the capabilities of their models. Other AI researchers tend to agree.

Nathan Lambert, an AI researcher with the non-profit AI2, compares AI reasoning modes to airplanes in a blog post. Both, he says, are manmade systems inspired by nature — human reasoning and bird flight, respectively — but they operate through entirely different mechanisms. That doesn’t make them any less useful, or any less capable of achieving similar outcomes.

A group of AI researchers from OpenAI, Anthropic, and Google DeepMind agreed in a recent position paper that AI reasoning models are not well understood today, and more research is needed. It may be too early to confidently claim what exactly is going on inside them.

The AI agents on the market today work best for well-defined, verifiable domains such as coding. OpenAI’s Codex agent aims to help software engineers offload simple coding tasks. Meanwhile, Anthropic’s models have become particularly popular in AI coding tools like Cursor and Claude Code — these are some of the first AI agents that people are willing to pay up for.

However, general purpose AI agents like OpenAI’s ChatGPT Agent and Perplexity’s Comet struggle with many of the complex, subjective tasks people want to automate. When trying to use these tools for online shopping or finding a long-term parking spot, I’ve found the agents take longer than I’d like and make silly mistakes.

Agents are, of course, early systems that will undoubtedly improve. But researchers must first figure out how to better train the underlying models to complete tasks that are more subjective.

“Like many problems in machine learning, it’s a data problem,” said Lightman, when asked about the limitations of agents on subjective tasks. “Some of the research I’m really excited about right now is figuring out how to train on less verifiable tasks. We have some leads on how to do these things.”

Noam Brown, an OpenAI researcher who helped create the IMO model and o1, told TechCrunch that OpenAI has new general-purpose RL techniques which allow them to teach AI models skills that aren’t easily verified. This was how the company built the model which achieved a gold medal at IMO, he said.

OpenAI’s IMO model was a newer AI system that spawns multiple agents, which then simultaneously explore several ideas, and then choose the best possible answer. These types of AI models are becoming more popular; Google and xAI have recently released state-of-the-art models using this technique.

“I think these models will become more capable at math, and I think they’ll get more capable in other reasoning areas as well,” said Brown. “The progress has been incredibly fast. I don’t see any reason to think it will slow down.”

These techniques may help OpenAI’s models become more performant, gains that could show up in the company’s upcoming GPT-5 model. OpenAI hopes to assert its dominance over competitors with the launch of GPT-5, ideally offering the best AI model to power agents for developers and consumers.

But the company also wants to make its products simpler to use. El Kishky says OpenAI wants to develop AI agents that intuitively understand what users want, without requiring them to select specific settings. He says OpenAI aims to build AI systems that understand when to call up certain tools, and how long to reason for.

These ideas paint a picture of an ultimate version of ChatGPT: an agent that can do anything on the internet for you, and understand how you want it to be done. That’s a much different product than what ChatGPT is today, but the company’s research is squarely headed in this direction.

While OpenAI undoubtedly led the AI industry a few years ago, the company now faces a tranche of worthy opponents. The question is no longer just whether OpenAI can deliver its agentic future, but can the company do so before Google, Anthropic, xAI, or Meta beat them to it?

A new law approved by the kingdom is set to give non-Saudis rights to buy homes in designated zones

Saudi Arabia could replicate the success of Dubai in attracting foreign property investors, including those from Hong Kong and mainland China, following the kingdom’s relaxation of homebuying rules, according to experts.

A new law approved by the kingdom last month is set to allow non-Saudis to buy real estate in designated zones. The law will take effect in January.

“Saudi Arabia can absolutely succeed at making itself a major destination for investment and residency,” said Kashif Ansari, co-founder and CEO of real estate broker Juwai IQI. “The new law will give non-resident foreigners a clear path to own property in Saudi Arabia for the first time.”

Ansari said the Saudi capital of Riyadh and another major city, Jeddah, offer greater value than Dubai and Abu Dhabi in the United Arab Emirates (UAE). For comparably sized residential units, the average price per square foot in Jeddah is US$100 and in Riyadh, it is US$123. In Dubai, that figure rises to US$400.

Foreigners owned around 43 per cent of all homes in Dubai, according to data compiled by the think tank EU Tax Observatory. Sales of homes in Dubai worth more than 10 million dirhams (US$2.7 million) jumped tenfold to 4,670 in 2024 from 2020, according to data tracked by Savills. In the first quarter of this year, 1,300 such homes changed hands, up 31 per cent from a year earlier, the property consultancy added.

Chinese buyers accounted for 14 per cent of Dubai property sales so far this year, up from 13 per cent a year earlier, Savills said.

Meanwhile, Hong Kong and the UK were the top sources of ultra-wealthy individuals who owned a second home in Abu Dhabi, according to a study released last month by US-based Altrata, which tracks wealthy individuals.

Saudi Arabia’s property market was projected to nearly double in size from now to US$133 billion by 2033, according to Patricia Casaburi, CEO of Global Citizen Solutions, a boutique investment migration consultancy with offices in the UK, Portugal and Brazil.

“What really catches my attention are the rental yields,” Casaburi said. “At a 6.75 per cent average and with Riyadh hitting nearly 9 per cent, these returns significantly outperform most developed markets. Combine that with zero personal income tax, no capital gains tax and no inheritance tax and the net returns become even more attractive.”

Casaburi likened the kingdom’s plan to relax rules surrounding foreign property ownership to the evolution of Dubai’s market when it opened to international buyers. Liquidity and accessibility in the kingdom should increase, she said.

Under the Saudi Vision 2030 plan, the country is modernising and diversifying its economy to reduce its reliance on oil. The initiative included mega infrastructure projects such as The Line, a glass-panelled development billed as the crown jewel of Neom, a city that is being built in Tabuk province.

“Since the launch of the [plan], the real estate market in Saudi Arabia has grown noticeably,” said Jihad Elabbas, senior client adviser in the Middle East and North Africa at migration consultancy Henley & Partners.

“Heavy investment in infrastructure has been a key factor, including the new metro service in the capital and the upcoming opening of King Salman Park, which is said to be the largest urban park in the world,” he said.

Juwai IQI’s Ansari said Muslims were the most likely to invest, noting “significant pent-up demand among high-net-worth” individuals.

Investors based in Hong Kong and mainland China who have business interests in the kingdom might also be interested in buying homes there, Ansari said.

“Chinese companies are actively engaged in projects across Saudi Arabia and the local Chinese expat population is growing,” he said.

Between 2005 and June 2024, China invested US$53.85 billion in infrastructure projects in Saudi Arabia, according to data cited in January by the Carnegie Foundation, a US-based think tank.

Ansari said Saudi Arabia could further relax residency rules to attract foreign investors.

“Riyadh can unlock demand at scale from Greater China with more inviting residency rules,” Ansari said. “Crucially, Beijing and Hong Kong buyers will compare Saudi entry rules with Dubai’s. So, the two countries will be competing on their residency terms. With golden visa rules similar to those in the UAE, Saudi Arabia will be appealing to investors and to families looking for relocation or second-home status.”

Casaburi of Global Citizen said Dubai was appealing to lifestyle investors and those seeking established international communities, while Saudi Arabia was positioning itself to attract institutional money and family offices.

Saudi Arabia was moving steadily towards becoming a destination for wealthy families, said Elabbas of Henley.

“However, it is to be expected that the kingdom will take some time to catch up with its neighbour, the UAE,” he said.

Engineers propose a solution that could dramatically boost UAVs’ prospects of evading advanced air defences

In Ukraine, about nine in 10 Russian drones have been shot down by advanced air defences, according to various estimates.

Ukrainian air force data suggested that about 15 per cent of drones had penetrated its defences between April and June — rising from just 5 per cent previously.

But a group of Chinese aerospace engineers and defence researchers have now proposed a radical technological enhancement for combat drones that may dramatically increase their ability to survive to nearly 90 per cent.

At the heart of the proposal was an innovative concept: fitting compact, side-mounted rocket boosters to small or medium-sized drones so they can perform instantaneous, high-G manoeuvres in the final seconds before a missile impact.

According to the researchers, this “terminal evasion” system allowed drones to perform abrupt, unpredictable course changes that even the most sophisticated missiles could not track or follow.

In extensive digital simulations detailed in a paper published in the Chinese defence journal Acta Armamentarii last month, the system saw a huge improvement in survival rates, passing 87 per cent.

In many cases, the drones effectively caused missiles to detonate harmlessly in empty space.

In modern wars, including the conflict between Russia and Ukraine, combatants have “extensively employed drones for reconnaissance and aerial combat, making [them] increasingly crucial on the battlefield,” wrote the project team led by Bi Wenhao, an associate researcher with the National Key Laboratory of Aircraft Configuration Design in Northwestern Polytechnical University in Xian.

Chinese military analysts, after studying the war in Ukraine and other areas, “place higher demands on the evasion capability and survivability of unmanned combat aircraft,” the team wrote.

Traditionally, drones are expected to make evasive moves long before the missile hits, but this can end their mission. As an alternative, Bi’s team suggested taking evasive action at the last possible moment.

The concept hinges on three critical principles, the first is precise timing, which means the anti-drone missile must ignite within a one to two-second window before impact – early enough to alter trajectory, late enough to deny the missile time to correct.

The second is directional intelligence: the system must determine whether to climb, dive, or veer laterally based on the missile’s approach vector.

The last of the three is thrust: the boosters must generate at least 16Gs of acceleration – far beyond what conventional aerodynamic control surfaces can achieve – ensuring a sudden, disorienting shift in flight path.

The project team faced some formidable challenges in integrating the rockets into a drone’s airframe without disrupting aerodynamic stability.

The forces unleashed during ignition must be perfectly synchronised with the drone’s flight control surfaces and onboard software – otherwise, the violent thrust could cause the aircraft to spin out of control or disintegrate mid-air.

Moreover, the system operates within a razor-thin time window. The booster burns for only about two seconds. Any delay – by mere fractions of a second – could render the manoeuvre useless.

There are also trade-offs that have to be considered. Adding weight, consuming payload space, and reducing fuel capacity all threatened the drone’s range and mission endurance.

The researchers said they had validated their model through thousands of combat simulations and found that their evasion algorithm was able to execute complex compound manoeuvres to maximise the drones’ ability to evade defences.

Once successfully miniaturised and integrated, such evasion systems could transform inexpensive drones into formidable, resilient strike platforms – capable of surviving in airspace once considered too dangerous.

In a conflict such as the war in Ukraine, where drones have already proven pivotal in intelligence gathering and precision attacks, a survivability boost of this magnitude could dramatically alter the nature of aerial warfare.

However, the concept has yet to be tested on a real battlefield. The drone would need to see the threat coming, sometimes from more than 200km (120 miles) away, and requires support from an integrated space-air-ground surveillance network.

Other challenges to overcome include delayed data transmission, particularly from distant satellites; signal jamming and misinformation; and insufficient onboard computing power for split-second decision-making.

The technology was designed to evade missiles with a top speed of Mach 4. Its effectiveness against hypersonic weapons remains unknown.

Freight groups warn the $250bn megamerger could raise prices and cut services

US railroad customer groups have demanded regulators block or put onerous conditions on the proposed merger of Union Pacific and Norfolk Southern, as both sides prepare for a regulatory battle over the $250bn deal.

Seven associations of shippers — as freight railroads’ customers are known — have expressed concern that the planned deal announced on July 29 would significantly increase the power of the merged railroad to raise prices or reduce service standards.

Opponents of a deal are also concerned it could prompt a further merger between BNSF and CSX.

A merger between UP — the biggest US rail network — and NS — the fourth biggest — would create the first US railroad able to operate transcontinental services all the way from the Pacific to the Atlantic coasts.

Ann Warner, a lobbyist working on behalf of the Freight Rail Customer Alliance, the National Industrial Transportation League and the Private Railcar Food and Beverage Association, said shippers were “overwhelmingly” opposed to the proposed deal.

“Every shipper is expressing concern and all the shippers have a long-standing position against further consolidation in the railroad industry,” Warner said.

Another customer group, the Alliance for Chemical Distribution, called on regulator the Surface Transportation Board (STB) to block the deal.

Eric Byer, the group’s president, said approval of the deal would be contrary to the public interest.

“Approving a transcontinental megamerger will benefit the merging rail companies and Wall Street, at the expense of US chemical distribution companies who are critical contributors to the American economy,” Byer said.

The American Fuel and Petrochemical Manufacturers said its members had “great concerns” about further consolidation of the freight rail industry.

“We fear this latest announcement will only compound the service problems refining and petrochemical shippers already face,” it said.

The deal is contentious because the number of large — or Class I — railroads in the US and Canada has fallen from about 40 in 1980 to just six, including the two Canada-based operators Canadian National and Canadian Pacific Kansas City.

Among the four US Class Is, two — UP and BNSF — operate west of the Mississippi, while NS and CSX operate in the eastern US.

Shipper groups fear that a further reduction in competition would leave them with less choice about which network to use.

Some groups, including the Soy Transportation Coalition, have expressed concern that a UP-NS merger could prompt BNSF and CSX to merge, reducing choice even further.

Ancora Holdings, an activist group that last year targeted Norfolk Southern, has been building a stake in CSX, its chief executive has said.

UP and NS have argued that an approved merger would benefit shippers by eliminating interchanges between railroads in places such as Chicago and St Louis, which often become congested.

In response to the shippers’ concerns, UP said its combination with Norfolk Southern was “about improving outcomes for customers and enhancing the US supply chain”.

“We’ve already spoken with more than 100 customers who are excited about the prospect of faster, more accessible, sustainable and lower-cost rail options,” it said. “They recognise the value of having end-to-end visibility across a transcontinental network and competitive single-line pricing.”

The STB will consider the deal in a lengthy process expected to take about two years. The two parties have said it will be six months before they even submit their application to the board.

Henry Posner, a veteran railroad investor and chair of the Iowa Interstate Railroad, said he believed only a minority of shippers thought the advantages of having “slightly easier” journeys between the eastern and western US outweighed the disadvantages of reduced competition.

The American Chemistry Council has also expressed deep concerns about a potential transaction.

The Intermodal Association of North America (IANA), which represents groups moving shipping containers by rail, sounded more supportive, however. Hub Group, an intermodal logistics company, has also welcomed the planned merger.

Because they typically handle higher value, time-critical cargo, intermodal shippers are more concerned than others about hold-ups during changeovers between railroads.

In a statement, IANA said intermodal rail thrived when there were “strong efficiencies, a focus on growth and a commitment to customer service”. “As this merger moves forward, we will be looking for these core values to be reinforced,” it said.