'We don't just want to create technology, we want to have a positive impact on the world.' – Stefan Leichenauer, SandboxAQ
Stefan Leichenauer is a man on a mission.
He is driven by the fact that he works for a company that is committed to making the world a better place.
That company is SandboxAQ, a B2B company that delivers AI solutions that addresses some of the world's greatest challenges.
SandboxAQ was born out of Alphabet Inc. as an independent capital-backed company in 2022.
Over the last number of years, it has grown exponentially across multiple global markets, and has a major partnership here in the Middle East region with Aramco.
Leichenauer spoke to CNME, about why the company wants to deliver technologies that have a positive impact on society, and the critical role played by LQMs in enabling the transformation pf industries such as the Oil & Gas sector.
In a recent op-ed, the VP of Engineering at SandboxAQ made the case for enterprises to shift their focus away from LLMs and to look at the LQMs to foster real change across their organisation.
According to Leichenauer, LLMs have limitations, and in order to solve the really complex challenges facing the world then businesses need to start looking at LQMs.
'Firstly, let me say that I think LLMs are fantastic, and we are not working to get rid of them. However, LLMs can't do everything by themselves, and I think that's the point that I am making, and I think more and more people are starting to realise that LLMs have their limitations. If you look at the latest LLMs that have been released over the last 3 years, then it seems like every release has a new set of capabilities that can do so much more, but we have sort of hit a ceiling of late. If you examine the latest releases of Llama 4 and GPT4.5 they are only incrementally better than what has come previously. So, I think there has been a realisation that LLMs as a capability are great, processing text and generating images then it is fantastic, but there is a whole set of capabilities that LLMs are just not going to get to by themselves,' said Leichenauer.
The capabilities that LLMs are not going to be able to get to by themselves is associated with quantitative reasoning, and this is where LQMs come to the fore.
'LQMs is designed to model the physical with chemistry, physics, and medicine, and is essentially focused on doing things that has absolutely nothing to do with language-based content. You need other tools in the tool box, and that's where LQMs come in. LQMs are basically providing those other tools in toolbox and they compliment the capabilities provided by LLMs,' said Leichenauer.
In his op-ed, Leichenauer also claimed that when precision is paramount then LQMs are indispensable, and said momentum was beginning to swing in favour of LQMs.
'We're now seeing more proof points that LQMs. I think in the past people would have deployed LLMs on to any given problem to see what works, and what doesn't, and I think everyone has been doing proof of concept trials with LLMs, but they've fallen short for a couple of reasons. As I stated earlier, in some areas they are fantastic, but in other areas they have fallen short. One of the reasons for this is the fact that LLMs are very non-transparent in terms of their reasoning. LLMs will give you an answer, but why is it true? And the LLM could be hallucinating, and we know that's been a big problem in some areas. Hallucinations are fine when it comes to generating an image, maybe it has the wrong number of fingers, but when it comes to creating a new molecule for Aramco, that is designed to making their processing plants more efficient, then you can't get that wrong because that's going to cost you a billion dollars. You need your answer to be correct, you need it to be grounded in real understanding of the problem, and LQMs can provide that verifiability and transparency,' said Leichenauer.
As aforementioned above, SandboxAQ have enjoyed great success since spinning out of Alphabet Inc. in 2022, and are working with some of the biggest companies in the Middle East, including Aramco, who are the biggest integrated energy and chemicals company in the world.
He spoke about their partnership, and again reiterated their mission which is to build purposeful technology designed to improve society.
'Our goal at Sandbox at the end of the day is not to create technology, of course we love to create technology, but we are doing it for a purpose. Ultimately, our goal is to have a positive impact on the world, and it just so happens that LQM technology is a great way to have a positive impact. The impact areas that we care about the most such as the medicine, pharmaceuticals, medical devices and GPS free navigation is something that we are very passionate about. These are all powered by LQMs. In terms of our collaboration with Aramco, the oil & gas industry is a really important industry in the world. However, we are all acutely aware that as we move forward, we need to be better about being environmentally friendly, and more efficient with our energy and more sustainable. We need to always be looking at better techniques, and Aramco is a real leader and pioneer when it comes to these sort of techniques,' said Leichenauer.
He went into more detail in relation to how LQM technology is enabling Aramco to transform, and how the technology is helping the global energy incumbent to be more sustainable and efficient.
'Aramco is not an AI company, they are an oil & gas company, so we are here to help our partners like Aramco to advance their operations to be able to do things in a much better way. SandboxAQ provides software tools, AI models and the LQMs that really help them to transform the way they operate their business. What we're doing with Aramco specifically is partnering with them to look closer at the oil & gas processing facilities. Ultimately, a lot of what is happening there is you've essentially got liquids and gases flowing through pipes and going through various kinds of processes, refineries and machines. However, in order to make those processes more efficient, one way to make them more efficient is to model them computationally better,' said Leichenauer.
Leichenauer conceded that these processes are complex, but insisted that in order to make them more efficient and sustainable then companies like Aramco had to implement LQM technologies.
'It's a complex physical process, and if you want to make your plants more efficient, and reduce emissions and waste then modelling that process computationally allows you to make tweaks and changes virtually to enable you to implement them in real-life. Modelling all of those processes computationally is something that our software is helping Aramco with,' said Leichenauer.
Leichenauer is delighted at the progress SandboxAQ has made with Aramco since their collaboration started, and believes that by 2030, it will fundamentally be a completely different business.
'The part that Sandbox has control over, and the computational modelling that enables these kinds of changes, the good news is, well from our perspective anyway is relatively simply compared to actually implementing these things physically. We have been working with Aramco for several months now, and we've already achieved significant milestones with our modelling. The LQMs that can do that sort of modelling and give you the answers and the playbook that what you need to do to make the changes those exist, and in a matter of months we have made huge progress on that. If I had to speculate a little bit then I'd guess that in the next 5 years we'll see a lot more changes coming through and being implemented. It may take longer to become 100% sustainable and 100% green, but in the oil and gas industry and other industries we can affect real changes and see real progress in a sort of 5-year timeline. By 2030 or so, a lot of the work we are doing today will have real tangible impact by then,' said Leichenauer.
Another industry that SandboxAQ is looking to transform in order to ensure they are having a meaningful impact on society is the healthcare industry.
'The healthcare sector is a major industry for us. It is a major source of grand challenges for the world, but we have seen a lot of progress in the last years in terms of how technology is being used to transform healthcare. When we are talking about real positive impact on the world then there's almost no better place to have that impact than in healthcare. Within healthcare, there is obviously the pharmaceutical industry, and there's always a lot to do in that space, and in terms of medical diagnostics that is a space that also can be transformed. The MRI machine is an amazing machine, it transformed medicine when it was invented several decades ago, but it is big, it is expensive, and it's clunky, and it takes a lot of expertise to use it. The next-generation of medical diagnostic devices can bring the kind of transformative impact of the MRI machine, but in a form factor that is more like an ultrasound machine, where it can something that can be much smaller and can be in every hospital emergency room. That kind of technology is coming, and some of that is what we are working on and using LQMs to enable,' said Leichenauer.
Leichenauer outlined that SandboxAQ is working on a diagnostic designed to tackle the issue of heart disease.
'We're working on a device right now using LQMs that is specifically for diagnosing heart disease, and various kinds of heart disease in an emergency room setting in a way that you could actually apply it to every patient that walks in complaining of heart problems, or persistent heart pain. One of the first things that you do is take five minutes to give them a scan using the machine, and that really improves the care of the patient, and heart disease is one of the biggest killers in the world, so this is a truly transformative device. At the minute, we have a prototype device being tested in hospitals right now, and within a couple of years I'd expect this device to be used on a everyday basis in hospitals. Early indications of the prototype is that we are on the right track, and appear to be doing a good job. However, you have to prove you're doing a good job and pass regulations and so on before you can actually go to market with such a device, but the technology is there and we are actively working on it,' said Leichenauer.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Tahawul Tech
10-07-2025
- Tahawul Tech
'We can no longer afford to build AI on brittle foundations past their prime.' – Haider Aziz, VAST Data
Haider Aziz, Managing Director, META at VAST Data, has penned a compelling op-ed exclusively for in which he highlights that vector databases designed to serve as the connective tissue between LLMs and real-time enterprise knowledge are no longer fit for purpose. Across the Gulf, we're witnessing a once-in-a-generation transformation. From Abu Dhabi's partnership with OpenAI on the Stargate initiative to the continued momentum of homegrown large language models like Falcon and Jais, and more recently the news that an AI system will become an official advisory member of the UAE Cabinet from 2026, the region isn't just adopting AI, it's actively shaping its future. Government ministries have appointed AI leads, sovereign funds are underwriting compute capacity at national scale, and public-private collaborations are redefining what's possible across energy, healthcare, and finance. In this context, AI infrastructure isn't just a technical consideration. It's becoming national infrastructure. But for all the progress in model development and GPU deployment, one critical component is quietly falling behind: the retrieval layer that underpins modern AI systems. More specifically, the vector databases meant to serve as the connective tissue between large language models and real-time enterprise knowledge. As it stands, they just aren't fit for purpose. From Prototypes to Production In the early days of Retrieval-Augmented Generation (RAG), use cases were modest. Enterprises indexed a few internal PDFs, connected them to an LLM, and deployed chatbots or semantic search tools with impressive early results. These pilots proved the potential of retrieval to ground AI outputs in enterprise-specific context. But today, the demands are very different. Organisations across sectors, from government to financial services to energy, are building AI agents to operate across petabytes of sensitive, constantly changing data. The Middle East, in fact, may have an advantage here: with less legacy infrastructure than many mature markets, there's greater freedom to build from first principles. But even in forward-leaning environments, retrieval at scale introduces real architectural complexity. And globally, this is where the stress begins to show. The shift from controlled pilots to production-scale systems has exposed the limitations of retrieval stacks not designed for real-time indexing, fine-grained access control, or trillion-vector search performance. What worked in a lab no longer works in the enterprise. We've proven that retrieval can work in contained, curated environments: the toy problem. But building systems that work across live, complex, permissioned enterprise data? That's the real challenge. When Vector Systems Become the Bottleneck Most vector databases were not designed to handle the size or complexity of modern enterprise data. In many current setups, data is embedded and indexed in batches—sometimes hourly, sometimes daily. This creates the impression of a responsive system, but in reality, it often serves answers based on outdated information. For industries like energy, finance, or healthcare, where conditions and decisions can shift by the second, this delay matters. A lot. Access control is another area where today's retrieval systems fall short. Enterprise environments often have strict policies around data permissions and auditing. Yet many retrieval pipelines rely on separate systems to handle those permissions. This means access rules are not always enforced at the moment of the query, and that creates a risk of exposing information that a user shouldn't see. Then there is the issue of scale. While some global AI developers have built systems capable of indexing the public internet, most enterprises already manage hundreds of petabytes of internal data. Few of today's vector tools were built to support that kind of volume, especially when you also need speed, accuracy, and secure access. These challenges all come into play as organisations shift from pilot projects to full production AI systems. When AI moves from the edge of the business to the centre, the expectation is not just that it works, but that it thinks. And thinking, in this context, depends on what the AI can reach. If the AI can only access a small, handpicked set of documents, its answers will be limited. Worse, when questions go beyond that narrow dataset, the model starts to guess, filling in gaps with incorrect or misleading information. This erodes confidence. By contrast, when every contract, meeting transcript, design spec, and operational log is embedded and searchable, the AI can draw from the full depth of institutional knowledge. Only with this kind of access can it uncover new insights, connect patterns, and offer guidance that feels relevant and trustworthy. But access alone isn't enough. The system also needs to know who is asking. Every AI response must be tied to the identity of the user making the request. That user should only see what they are authorised to see, nothing more. This is why it is so important for vector-based retrieval to follow the same access control policies as the source systems. It must enforce those rules automatically, in real time, for every single query. That is the foundation of trust. And it is one of the most critical elements in making AI enterprise-ready. Regional Ambition Requires Architectural Rethinking The UAE's AI strategy is not incremental. From the work of G42 across genomics, space, and sovereign clouds, to the AI-driven automation efforts in NEOM and Saudi Aramco, regional leaders are setting the bar high. These aren't controlled experiments, they are production-scale deployments that demand architectural maturity. To meet this ambition, we need to move beyond the previous infrastructure playbook as follows: Retrieval must be embedded within the data platform itself, not bolted on as a secondary service. Vector indexing must operate in real time, not on batch timers. Permissioning must be enforced atomically, from the data source through to the AI output. Retrieval systems must be capable of managing trillions of vectors, not just curated datasets. The good news is that this is achievable, but it requires moving past the prototype phase and rethinking how data infrastructure is built for the age of AI. What Comes Next In a region that is investing billions in sovereign AI capabilities, the retrieval layer cannot remain an afterthought. As more government and enterprise teams in the UAE and across the GCC shift from proof-of-concept to production deployment, the limitations of today's vector systems will become more apparent. The question is not whether retrieval will scale, it's whether organisations are ready to scale with it. To get there, we need infrastructure that treats vectors not as sidecar metadata, but as core intelligence, integrated, permission-aware, real-time, and built for the scale of national ambition. As many of you will know, in this part of the world, AI isn't just another tool in the stack. It's a strategic priority for the region. And, we can no longer afford to build it on brittle foundations past their prime.
Channel Post MEA
09-07-2025
- Channel Post MEA
Cybercriminals Increasingly Exploit AI Tools To Enhance Attacks: Cisco Talos
Cisco Talos has published a new report revealing how cybercriminals are increasingly abusing artificial intelligence (AI) tools – particularly large language models (LLMs) – to enhance their operations and evade traditional defenses. The findings highlight how both custom-built and jailbroken (modified) versions of LLMs are being used to generate malicious content at scale, signaling a new chapter in the cyber threat landscape. The report explores how threat actors are bypassing built-in safeguards legitimate AI tools use, creating harmful alternatives that cater to criminal demands. These unregulated models can produce phishing emails, malware, viruses and even assist in scanning websites for vulnerabilities. Some LLMs are being connected to external tools such as email accounts, credit card checkers, and more to streamline and amplify attack chains. Commenting on the report's findings, Fady Younes, Managing Director for Cybersecurity at Cisco Middle East, Africa, Türkiye, Romania and CIS, stated: 'While large language models offer enormous potential for innovation, they are also being weaponized by cybercriminals to scale and refine their attacks. This research highlights the critical need for AI governance, user vigilance, and foundational cybersecurity controls. By understanding how these tools are being exploited, organizations can better anticipate threats and reinforce their defenses accordingly. With recent innovations like Cisco AI Defense, we are committed to helping enterprises harness end-to-end protection as they build, use, and innovate with AI.' Cisco Talos researchers documented the emergence of malicious LLMs on underground forums, including names such as FraudGPT, DarkGPT, and WhiteRabbitNeo. These tools are advertised with features like phishing kit generation and ransomware creation, alongside card verification services. Interestingly, even the criminal ecosystem is not without its pitfalls – many so-called 'AI tools' are also scams targeting fellow cybercriminals. Beyond harmful models, attackers are also jailbreaking legitimate AI platforms using increasingly sophisticated techniques. These jailbreaks aim to bypass safety guardrails and alignment training to produce responses that would normally be blocked. The report also warns that LLMs themselves are becoming targets, as attackers are inserting backdoors into downloadable AI models to function as per the attacker's programming when activated. As a result, models using external data sources to find information are exposed to risks if threat actors tamper with the sources. Cisco Talos' findings underscore the dual nature of emerging technologies – offering powerful benefits but also introducing new vulnerabilities. As AI becomes more commonplace for enterprises and consumer systems, it is essential that security measures evolve in parallel. This includes scanning for tampered models, validating data sources, monitoring abnormal LLM behavior, and educating users on the risks of prompt manipulation. Cisco Talos continues to lead the global cybersecurity community by sharing actionable intelligence and insights. The full report, Cybercriminal Abuse of Large Language Models, is available at
The National
02-07-2025
- The National
Increased GPS jamming sparks interest in alternative navigation technology
One of the side effects of the hostilities between Israel and Iran has been the increased inaccuracies of navigation systems relied on by ships and, to a lesser extent, smartphone users in the Middle East. Most experts say those inaccuracies are probably the result of Global Positioning Satellite spoofing, known as GPS jamming. Israel, Iran, proxy groups, or all of the above might be to blame. Companies trying to protect their own interests might also be a contributing factor. Regardless, GPS jamming can be used offensively or defensively, and is often implemented by militaries to disrupt signals sent by navigation and location-based systems to gain a strategic advantage, avoid missiles or confuse the enemy. According to Jack Hidary, chief executive at SandboxAQ, a developer of products involving quantum technology and artificial intelligence, the very things that make GPS so accessible are the same things that make it so vulnerable. "It's a publicly registered and publicly available frequency that is widely accessible," he said, explaining that those attributes make GPS easily adaptable to various devices, operating systems and vast technology standards. "That also means that it's brittle by design," Mr Hidary added. A quick internet search shows how vulnerable GPS can be, with a plethora of devices available to just about anyone who wants them ranging from $130 to $2,000. That sort of easy access, combined with hostilities and geopolitical tension, have caused GPS headaches for shipping and even air. Amid the conflict between Israel and Iran, according to maritime insights and data provider Windward, about "970 ships per day" experienced GPS jamming in the Arabian Gulf and Strait of Hormuz. All that jamming can cause delays and be costly and, in turn, many companies are now looking into alternatives to GPS, with technology firms actively researching what works best. GPS alternatives? Over at SandboxAQ, Mr Hidary said the company is attracting a lot of interest in its proprietary AQNav technology from companies looking for reliable alternatives to GPS. He described AQNav as having three main components, a quantum sensor that picks up the magnetic field of the Earth, a graphics processing unit (GPU) and special AI software. Though it might seem complex at first, he said AQNav is rooted in and inspired by nature. 'Birds use it, literally," he explained, referring to the Earth's magnetic field and how its used by animals. "Billions of birds use it every day to go and do their business of migrating up and down this planet, and whales are using it too." Each square metre of Earth, Mr Hidary said, has a unique fingerprint that corresponds with Earth's magnetic field, and he said SandboxAQ's technology is able to make the most of that existing data. Perhaps most importantly is that it depends on far fewer elements, unlike GPS, Mr Hidary said. "It does not connect to a satellite, it does not connect to the cloud, it does not connect to the internet," he said. "Therefore there's no way to hack it." A number of companies are working on GPS alternatives. Some are experimenting with the use of low-Earth-orbit satellites, while others have pursued something called cellular network positioning, with mixed results. Nonetheless, the pursuit of more secure and therefore less fragile navigation systems is likely to continue, with interest at an all-time high due to geopolitical tension and GPS spoofing. In June, The European Union Aviation Safety Agency and the International Air Transport Association tentatively agreed to a comprehensive plan that would address GPS jamming. Some parts of the plan involve tightening export controls on GPS jamming devices, as well as maintaining back-up systems to create a "minimum operational network of traditional navigation aids". Tech experts and entrepreneurs like Mr Hidary say the pursuit of more navigation options and the improvement of GPS could pay big dividends for everyone. "We all depend on reliable transport, either for ourselves travelling and/or for cargo to come in, and inherently we're all dependent on a strong navigation system to make that happen," he said.
