Confidential/interoperable software enabling users to deploy AI in a private, controllable, and auditable way. 


New technologies and market pressures are pushing enterprises to redefine their IT architecture. For example, companies are trending towards cloud repatriation and multi-cloud solutions1. These strategic shifts are largely driven by cost and performance optimization to manage the future demand for data and AI. How do organisations maintain and control data and IP confidentiality in this new diverse landscape? New variables that will need to be considered are:

  1. Security and confidentiality of new AI products and services
    • AI privacy will require controls at the pre-, in-processing and deployment stages2
  2. Confidence in the data infrastructure that powers the new data economy and AI growth
    • Data will grow exponentially and will be mostly unstructured from a multitude of new data sources, exacerbating data privacy concerns3
  3. New ways that AI will be implemented (i.e. IoT/edge devices and edge data centers)
    • Edge investments will increase by 18% CAGR, leading to questions of how each new IoT device and the supporting data infrastructure will be kept confidential.4
  4. New regulation and expansion of existing like GDPR (i.e. new EU AI regulation)
    • Companies will need audits, data protocols and AI monitoring.5 

What is clear is that the future development and implementation of AI will require enterprises to rebalance their strategy. They must weigh the risk/reward between maintaining privacy/confidentiality to gain public/regulatory trust while developing the most competitive AI solution.

Data breaches can be very costly. Alongside the brand damage, regulators have shown they are willing to fine companies. One of the most notable incidents, Facebook/Cambridge Analytica scandal, cost  Meta $5B in fines. With AI becoming ubiquitous, the control of data use and confidentiality is critical. In the new EU AI regulation guidance, new higher thresholds are being set for potential fines, indicative of the increased need for rigour in data deployment1.

The vision of encloud is to create a foundation for the future AI revolution. We offer a data environment where enterprises can focus on and enhance their core operations without concerns about data privacy. With the increased focus on cost, performance, and operational optimization (“FinOps”), encloud can become a key strategic option to ensure high-value data is private, compliant, and deployment is scalable on any chosen IT architecture. At the core of our offering is confidential computing, we believe that confidential computing will become the de facto standard for sensitive data deployment, as confidential computing uniquely offers flexibility of deployment and robust privacy assurance.

What is encloud 

encloud is a confidential and interoperable software that enables organisations to deploy machine learning and AI like LLMs in a private, controllable and auditable way. It enables inference and training of AI in a privacy-assured runtime environment so that organisations can deploy, share and collaborate on their data easily while maintaining data confidentiality and IP of each party. This is achieved through confidential computing (TEE/secure enclaves), best-in-class encryption, and attestations – with AMD SEV and Nvidia H100 as core in our current and future offering.

At encloud, we believe existing practice around privacy and security will prove itself to be unfit for the modern data economy. We have designed our solution to fit together seamlessly to ensure true end-to-end privacy. In our solution we have been guided by best practices around data security and privacy like:   

  1. Guarantee privacy over the entire data lifecycle. 
  2. Govern data access  
  3. Prevent IP/AI algorithm leakage
  4. Audit workflows and deployment.

encloud’s solution improves on current privacy standards. For data pre-processing, we use envelope encryption to ensure input data cannot be discovered. encloud uses unique data identifiers as a strong audit mechanism ensuring the right data is deployed at the right time and in a permissioned environment. Finally, data is only decrypted and deployed in a trusted execution environment where no vendor has access to data, ensuring imperfect privacy measures like anonymisation and private networks are not required. For a fuller discussion on possible attack vectors and the privacy improvements made by confidential computing please see our recent blog. When workloads are deployed data is protected and isolated from attack and workflows are auditable through attestation.

Since AI algorithms can understand unstructured information, they are increasingly deployed in the real world where the data resides and in real time. This is not possible in a centralised cloud due to issues related to latency, bandwidth and privacy. However, private cloud solutions contain data locally, uploading analysis and insights to the cloud. The current privacy standard is for data to be anonymised, but this risks IP and data leakage when uploaded for model training purposes and in repeated workloads.  Private cloud thus safely empowers businesses in mission-critical areas – including high performance computing and AI – only when combined with secure and private solutions like encloud.

Organisations that gain most value by using encloud

Mid-sized businesses

encloud best serves organisations and industries that are looking to deploy their high-value data, for example in combination with public AI models. Typically, our target customers are mid tier companies for whom the safe deployment of data is problematic (cost, operational complexity etc). These companies may also lack vertical integration or have significant dependencies across their value chain. 

Historically, businesses would default to on-prem private networks to ensure their high-value data and IP are safe. However, mid-sized companies do not have appropriate scale or sophistication to manage in-house hardware infrastructure and face evolving IT security demands as they advance digital experience and enable disparate remote operations. They do not have the IT expertise to build the necessary privacy capabilities to ensure data is confidential and/or do not have the reputation to engender trust where needed. 

The complexity of deployment of this kind of hybrid solutions means consultants and system integrators have a vital role to play recommending and then implementing these solutions. 

encloud can help solve privacy for your clients

For clients seeking to optimise their compute infrastructure for their AI deployment, consultants can offer encloud as a simple solution that will fit into any existing IT infrastructure, on the cloud or on prem. Additionally, encloud will enhance existing IT architecture by enabling data collaboration across silos and/or different organisations. Accessing contextual data across these boundaries will elevate the performance of the next generation of AI products and services. Each data owner will have a technology-based assurance that their data is not manipulated, misused or exposed. As a technology solution that inherently guards confidential information, encloud makes data compliance and security easier – making sharing data simple with minimal need for trust and with auditability of how and where data is used.

Appendix: Example Use Cases with encloud

AI has seen adoption in multiple industries and we highlight some example use cases possible with encloud..

Financial Services

The financial services industry has witnessed widespread adoption of AI, with the Bank of England and the UK’s Financial Conduct Authority released publication “Machine learning in UK financial services” identifying that two thirds of respondents had already deployed machine learning in their businesses as of 2019.

Credit Score

By leveraging data across diverse sources and using AI, financial services can uncover latent features that determine creditworthiness – unlocking opportunities to lend to overlooked customers or identify risks. Using encloud, financial services will be able to access new data sources not previously possible..

Fraud Detection

Sophisticated fraud often leaves evidence across many different data sources. Using AI that spans across these data sources can help spot suspicious behaviour. With encloud, an AI can be deployed to investigate data from various sources enabled by private sharing. Only pre-approved operations can be performed, which ensures the AI does not expose commercially sensitive or personal data.

Consumer Packaged Goods

Away from financial services, Consumer Packaged Goods (CPG) companies, for example, have already deployed AI successfully which, together with advanced analytics deployment, have led to revenue, productivity, and marketing expenditure improvements. Meanwhile, 83% of retailers and CPG firms formerly surveyed declared that AI would become a “mainstream technology” for them in 2021, with the same survey reporting that current benefits for those already deploying AI include improved customer experience, enhanced employee upskilling, improved decision-making, and risk reduction. AI provides CPG firms with benefits including accurate forecasting, improved supply chain management and enhanced targeting of scarce resources.

Supply Chain Management

A robust supply chain management, for example at Consumer Packaged Goods (CPG) companies, requires predictive capabilities for the entire supply chain – from raw material to the end customer. AI has shown potential to identify market needs and predict the need for preventative actions. However, for an AI to fully manage the supply chain and minimise risks, it will need access to data across the supply chain. encloud enables the sharing of data from suppliers without exposing their data or IP increasing transparency.

  7. 1.