Deploy AI Analytics Where Your Data Lives

AI experiments are rapidly moving into organization-wide production. But in many cases, the one-size-fits-all approach of cloud-only AI with vendor-managed models isn’t viable.

When AI becomes part of everyday business processes, organizations face questions often avoided during earlier experimentation: Where does the application run? Which models can it use? Where is sensitive data processed? Does the architecture meet existing security, regulatory, and data sovereignty requirements?

New research from BARC and Actian shows how deployment requirements are changing as AI moves into production. The Where Enterprise AI Runs report found that three in four organizations use cloud services for AI workloads in some form, yet 76% say the importance of deploying AI outside the cloud has increased over the past 12 months. The shift reflects growing requirements around sovereignty, sensitive information, auditability, provider dependency, resilience, and business criticality.

That doesn’t mean enterprises are abandoning the cloud. But it does reflect their need for choice over where individual AI workloads run.

Actian AI Analyst now supports self-hosted deployment and bring-your-own LLM (BYOLLM), giving organizations more control over where their AI analytics runs and which models they use.

Trusted AI Analytics, With More Deployment Choice

AI Analyst gives business users an AI data analyst for enterprise data. They can ask questions in natural language, investigate changes, create reports, and monitor the business without waiting for a new dashboard or submitting every question to the analytics team.

Behind those interactions, AI Analyst grounds analysis in governed metrics, dimensions, relationships, and business definitions. The logic used to produce results remains visible, so teams can validate how an answer was reached rather than simply accepting an AI-generated response.

Until now, AI Analyst has been delivered as managed SaaS. That remains the simplest option for many organizations, but it doesn’t fit every production environment.

A financial institution may require an analytics application to run within its private cloud alongside other approved AI services. A manufacturer may need sensitive operational data to remain within a particular environment. A healthcare provider might already have approved LLM providers and model governance policies that every new AI application needs to follow.

These requirements become harder to work around as AI moves from experimentation into production. BARC describes the emerging approach as deliberate workload placement. Deciding where AI should run based on the control, risk, performance, and infrastructure requirements of each workload.

Run AI Analyst Where Your Requirements Demand

Self-hosted deployment lets organizations run AI Analyst within infrastructure they control rather than exclusively using Actian’s cloud environment.

For organizations with security, regulatory, data sovereignty, or architectural requirements around where applications operate, AI Analyst can fit within the environment they’ve already established. That could mean deploying in a private cloud alongside existing data and AI services or keeping the application on-premises when workloads need to remain within the organization’s own environment.

The choice comes down to the requirements of the workload. Use managed SaaS when you want Actian to manage the application infrastructure. Self-host AI Analyst when security, regulatory, sovereignty, or architectural requirements call for the application to run within infrastructure you control.

Deployment model choice doesn’t change the product experience. Self-hosted AI Analyst provides the same functionality and performance as the cloud version, including the same governed semantic foundation and visibility into how results are produced.

Use the LLM That Fits Your AI Strategy

Where the application runs is only one part of the equation. Many enterprises have also established their own relationships with LLM providers, AI gateways, and cloud AI services.

Bring your own LLM lets organizations connect AI Analyst to their existing LLM provider or model endpoint rather than relying on an Actian-hosted model. This gives teams more control over model choice and lets AI Analyst fit into existing enterprise AI infrastructure and governance policies.

That can include a customer-managed account with a commercial LLM provider such as OpenAI or Anthropic, or models accessed through an enterprise AI gateway or cloud AI service such as AWS Bedrock, Google Vertex AI, or Azure AI, depending on the customer’s architecture and authentication requirements.

For some organizations, BYOLLM is all the additional control they need. Others may have model infrastructure that is only accessible from within their private network. In those environments, self-hosting AI Analyst provides another level of deployment control while preserving the same analytics functionality.

As AI moves from experimentation into everyday business use, organizations need the flexibility to fit AI into the infrastructure, security requirements, and model strategy they already have. With self-hosted deployment and BYOLLM, AI Analyst can now fit the architecture and AI strategy they’ve chosen.

Start a free trial of Actian AI Analyst to see how governed AI analytics works with your data.

About Author

About Andrew Petersen

Andrew leads product marketing for Actian's AI portfolio. With a background spanning product marketing, customer research, content strategy, and journalism, Andrew focuses on positioning complex data and AI technologies to connect with tangible business outcomes. Prior to Actian, he held product marketing roles at CData, data.world, and Workato.