Integrated Quantum Technologies’ flagship product VEIL enables enterprises to secure data before it enters ML models using a new privacy-preserving framework that removes personally identifiable information (PII), while conserving and enhancing data utility. VEIL allows for scalable innovation without tradeoffs, and is designed to withstand future quantum computing threats. The company focuses on enhancing data security, regulatory compliance, and performance for AI/ML systems. Pulse 2.0 interviewed Integrated Quantum Technologies CTO Anita Oehley to learn more.
Challenges Enterprises Face Today
Having spent 25 years transforming enterprise-scale organizations including leading organizations like AWS Marketplace, what are the biggest challenges enterprises are facing today? Oehley said:
“One of the biggest challenges enterprises are facing today is how to balance innovation with risk. There is strong pressure to scale AI quickly, but many organizations do not have the right approaches to protect sensitive data.”
“As enterprises continue to rely on machine learning to power their core systems, traditional security tools fall short. They need to use methods designed for ML models and modern operating environments. Machine learning models depend on large volumes of data including sensitive data to be as effective as possible and this makes these systems a prime target for adversaries.”
How These Challenges Are Being Addressed
How is Integrated Quantum Technologies, where you’re now CTO, addressing these challenges? Oehley shared:
“Integrated Quantum Technologies is addressing these challenges at the data layer, by building security into the foundation of modern AI infrastructure rather than layering it on after the fact. Through our signature product, VEIL, we transform sensitive data into secure, non-invertible representations before it ever enters machine learning pipelines, minimizing risk while still enabling models to perform effectively.”
“What is fundamentally different is that we don’t just protect data in the machine learning lifecycle, we make it efficient to use. VEIL applies a compression technique that significantly reduces the data size while preserving, and in some cases improving, model accuracy. In fact, we have been seeing improvements in model accuracy in every research we have conducted so far. This helps organizations scale their AI/ML solutions much more efficiently, reducing ML infrastructure costs and accelerating time to market.”
“At the same time, VEIL is quantum-resilient by design. Because data is non-invertible, it cannot be reconstructed, which changes the risk profile significantly for organizations with sensitive data. This enables enterprises to scale machine learning models in a way that is both secure and operationally efficient, without forcing a trade-off between model performance and data exposure.”
Jeremy Samuelson, EVP AI and Innovation at Integrated Quantum Technologies, discussing how VEIL works
Inflection Point For Post-Quantum And AI Security
Why is now the inflection point for post-quantum and AI security — what has changed in the last 12–24 months? Oehley pointed out:
“Over the last 12-24 months, the inflection point hasn’t been about the introduction of Machine Learning. Some organizations have been using machine learning for a few years now. What’s changed is how these models are being operationalized at scale. As ML has moved from contained and research-driven use cases into production systems, the volume of sensitive data being actively used has increased significantly, along with the number of systems, users, and partners interacting with it. That’s the inflection point. Machine learning hasn’t changed, but the way they are embedded into business operations, and the level of exposure around them, has.”
Core Problem Enterprises Are Facing Today
What is the core problem enterprises are facing today that existing security solutions are not solving? Oehley noted:
“Enterprises do have strong security programs in place, but they often lack approaches specifically designed for how data is used inside machine learning. What’s then changed is how critical these models have become, since they now drive core business operations and decisions.”
“That’s the core problem: organizations are relying on methods that weren’t designed for this stage of the lifecycle, leaving a gap in how sensitive data is protected while it’s actually being used. VEIL addresses this problem at the source, protecting data before it enters the machine learning lifecycle.”
Building Post-Quantum AI Infrastructure
You’ve described IQT as building “post-quantum AI infrastructure.” How do you define this category, and why does it need to exist? Oehley explained:
“What we mean by ‘post-quantum AI infrastructure’ is building systems that address today’s machine learning security needs while also protecting against future risks as computational capabilities evolve, including quantum. Attacks are already in play with the most well-known one referred to as Harvest Now, Decrypt Later. This creates long-term exposure risk, especially for sensitive data used in machine learning. VEIL is designed with that in mind. It transforms data into non-invertible representations before it enters the machine learning pipelines; so, it reduces the risk of future decryption altogether. This category needs to exist because organizations need infrastructure that protects both how data is used today and how it could be exploited in the future.”
Differentiated Approach
What makes your approach fundamentally different from existing solutions like encryption, tokenization, or differential privacy? Oehley affirmed:
“The simplest way to explain it is that existing approaches, like encryption and tokenization and differential privacy, address different parts of the problem. They also come with trade-offs when applied to machine learning. Encryption and tokenization both protect data, but data typically needs to be a usable form for the ML model, and this creates exposure risks during processing. Differential privacy is designed for AI/ML and it introduces noise which can impact the model accuracy. There are other approaches like homomorphic encryption, but they can introduce significant performance and scalability challenges.”
“VEIL uses a different approach. It transforms the data into non-invertible representations that protect data in the machine learning pipeline. VEIL avoids computational overhead and performance trade-offs you typically see with other privacy-preserving techniques, making it lightweight and scalable.”
Enterprise Adoption
Where are you today in terms of deployment readiness, and what does enterprise adoption look like in practice? Oehley described:
“Our product is available commercially. We are currently working with early enterprise partners on pilot implementations. VEIL integrates into existing ML environments and that makes adopting VEIL practical.”
Feedback From Enterprise Customers
What are you hearing from enterprise customers — what is driving urgency for them right now? Oehley highlighted:
“What we are hearing from enterprise customers today is a growing tension between wanting to scale their AI/ML and being confident in how their data is being handled. On one side, there is increasing pressures from regulators and internal governance teams wanting control and visibility into how data is being used in their models. On the other hand, the business is pushing to move faster and embed these models into scalable production systems.”
“Organizations want to do both. They want to innovate fast and truly operationalize their AI/ML solutions, without taking on unnecessary data exposure risk. The stakes are higher now, both from a regulatory and reputational standpoint. So, the urgency isn’t just about security in isolation, it’s about being able to scale machine learning with confidence.”
Positioned Against Cybersecurity Vendors
How do you position yourselves against both traditional cybersecurity vendors and emerging AI security companies? Oehley pointed out:
“We don’t position ourselves as a cybersecurity solution. And we aren’t trying to replace one.”
“We operate at the data layer within AI/ML workflows, where data is actively used in the pipeline. That’s where most risk exists today. VEIL is designed to complement existing security tools. We focus on how data stays protected as it moves from ingestion to training and to inference.”
Impacting Cost And Scalability
How does your technology impact the cost and scalability of deploying AI at the enterprise level? Oehley emphasized:
“This is where customers see immediate value. VEIL compresses and transforms the data in the ML lifecycle. which has a direct impact on cost and scalability. Because data size is smaller and more efficient to work with, ML infrastructure costs go down, and moving data across environments becomes much easier. It also simplifies global deployments.”
“At the same time, organizations avoid the heavy compute burden you see with some other approaches, and they don’t need to create redundant environments just to manage risk.”
“So, it’s not just about reducing risk, it’s about making AI/ML more efficient and economically viable to scale.”
Key Indicators
For investors evaluating this space, what are the key indicators they should be paying attention to? Oehley told me:
“Innovation that cannot be deployed safely at scale ultimately creates more risk than value. If you’re looking at this space, focus on where AI/ML is moving into production, not just where there is experimentation and hype. That’s where the real challenge lies. Also, paying attention to regulatory pressures is key, as that impacts how systems are built and scaled. The companies that can handle both scalability and security are the ones to watch.”
Risks
If enterprises do not rethink how they secure AI infrastructure today, what is the risk moving forward? Oehley concluded:
“AI increases both opportunity and risk exposure. If the underlying infrastructure isn’t designed for that, small gaps turn into systemic issues very quickly. Over time, that erodes trust and increases cost and over time, it limits how far organizations can scale. The companies that get this right move faster and more confidently than everyone else.”
