Pro-Data has launched a sovereign AI platform for business

Cloud provider Pro-Data has launched a sovereign AI platform for businesses. Companies can use the platform to deploy ready-made AI models, test them, and integrate them into their applications via API, as well as develop, train, and fine-tune their own AI solutions using the platform's GPU resources.

The servers are located in Uzbekistan and Kazakhstan; customer data is processed within the chosen country and does not cross national borders.

What is a sovereign cloud

A sovereign cloud is one where the entire infrastructure is physically located within the customer's country and subject to its laws. The difference between it and a foreign service lies not in the technology, but in the jurisdiction: when servers are located abroad, company data falls under the legal framework of another state, making it subject to requests, restrictions, and rulings that neither the customer nor the provider can influence.

This has become a critical issue for AI. AI models do not work with anonymized statistics but with the core elements of a business—such as customer inquiries, contracts, transactions, and call recordings. By sending such data to a foreign API, a company transfers it entirely and irreversibly into an external environment. A sovereign platform resolves this issue at the architectural level: training and inference take place on domestic infrastructure, data remains subject to local regulations, and support, contracts, and billing operate within the local legal framework.

What it consists of

AI Platform is a catalog of ready-made models, an experimentation environment, and a GPU-based setting for developing AI solutions. Tools for creating AI agents capable of handling multi-step tasks and interacting with external systems will be added to the platform later.

1. Inference Studio — a catalog of ready-to-use models, a Playground for experiments, and API access. It supports inference, streaming, function calling, structured output, and monitoring.
2. AI Lab — a development and training environment: Jupyter, VS Code, GPU containers, custom Docker environments, and ML frameworks. This is where models are fine-tuned on your own data.
3. Agent Studio — a visual agent builder with scenario templates and integration into corporate systems. This component will be released later.

Access formats for computing power

Computing resources are available in three formats, ranging from an experimental setup to a fully isolated environment. The on-demand mode operates on an hourly payment basis, allowing you to test models on GPUs without significant hardware investment, select the right configuration, and only then switch to a dedicated environment.

Tasks the platform helps solve

• Examples of solutions that can be implemented on the platform
• Support Assistant - responds to customer requests and finds answers in the company’s internal knowledge base
• Document processing - extracts data from accounts and primary documents, classifies incoming applications, reduces long texts to the essence
• Developer assistant - generates and refactors code, writes autotests on an isolated circuit where the source code does not go
• Speech analytics - recognizes speech in real time, transcribes calls and shows what customers ask most often
• Scoring model - assesses risks and creditworthiness without releasing client data outside the perimeter
• Batch processing - parses accumulated data sets and retrains models during off-peak hours at a more favorable price

Data, infrastructure, and certification

Physical server hosting in Uzbekistan and Kazakhstan is a mandatory requirement for sensitive industries. Data never leaves the country and is processed in compliance with local regulations. Proximity to the infrastructure further reduces latency, which is critical for voice-based scenarios and real-time transaction scoring.

Equipment is housed in Tier III data centers. We offer up to 99.9% SLA availability. The infrastructure is ISO 27001:2022 certified, with 24/7 technical support.