Xerpa: Enterprise Generative AI Assistant
Your data, your rules.
With Xerpa, make the most of your organisation's internal knowledge and automate the execution of tasks involving different data sources and corporate platforms.
What is Xerpa?
Xerpa is the corporate AI assistant developed by Xeridia, designed to work securely and in a controlled way with your company's internal information.
Xerpa provides simple, natural-language access to all of your company's knowledge, even when it's spread across multiple sources — whether documents, document management systems, or corporate applications (ERPs, CRMs, databases, etc). With Xerpa, you can easily configure and change the AI models your users work with (both local open-source models, to guarantee data sovereignty, and third-party frontier models). It also makes it easy to configure agents that can automatically carry out tasks involving multiple data sources and corporate tools, always with maximum security and full control over what gets executed.
Harness generative AI without compromising on data sovereignty, governance and security
Xerpa is deployed on the client's own infrastructure, whether on-premises or in the cloud. The ability to configure which AI model is used means you can switch between models without losing users' history (their prompts, conversations, etc.), allows you to control usage and consumption across different models, and lets you manage access security to the various information sources centrally and from a single point, rather than through each individual AI model provider. It also provides three layers of security: its own access control, users secure access to corporate platforms, and prompt-level security, which can be configured for the entire organisation.
Xerpa has been designed to meet the needs of companies for whom information is a core asset or competitive advantage, and who need a solution that guarantees data sovereignty.
Key features of Xerpa
Data sovereignty.
You decide whether your data leaves your environment or not.
Xerpa is deployed on the client's infrastructure (private, public or hybrid cloud) and can be configured so that no data ever leaves it.
AI governance.
You choose which AI model to use.
Local models such as open-weight (GLM 5.3, Qwen, DeepSeek, Mistral, etc.) or third-party models (ChatGPT, Claude, Gemini, etc.). You can switch models transparently for all your users, without them losing their conversation history or data.
Security.
Control across three independent layers.
Access-level, integration-level and prompt-level security.
Xerpa's agents
AI agents that carry out tasks for you.
Xerpa lets you configure different agents to carry out tasks across multiple data sources and tools.
Simply describe the task, and Xerpa will call on the configured agents to complete it. When a task involves sending or modifying data, it will request the user's prior authorisation, in line with the configuration in place.
Request.
Natural-language query.
Describe the task, provide the necessary context, and indicate the result you expect.
Orchestration.
Distribution of work among agents.
Xerpa analyses the request and determines which configured agent or agents should handle it.
Control.
Confirmation prior to execution.
Before sending or modifying any data, Xerpa asks the user for confirmation. In read-only mode, for queries, searches and summaries, it responds directly.
Some examples of use cases
What sort of organisations is Xerpa designed for?
FAQs
What is Xerpa?
Xerpa is the corporate generative AI assistant developed by Xeridia to work securely with your company's internal information. It answers questions in natural language about documents and systems such as databases, ERPs or CRMs, and carries out tasks on them via agents, subject to user authorisation. Its key features are data sovereignty, AI model governance and three-level security.
How can I ensure data sovereignty with Xerpa?
Xerpa is deployed in the client's own environment and can be configured to exclusively use open-source models running locally on their infrastructure. This means data never has to leave for third-party environments.
If I use Xerpa, will I still be able to use commercial AI models?
Yes, that's possible. Xerpa can be configured to use multiple models, allowing you to decide which one to use depending on the company's needs. The configurable models include those deployed locally (proprietary models, open-source, open-weight (GLM 5.3, Qwen, DeepSeek, Mistral, etc.)) as well as third-party models (ChatGPT, Gemini, Claude...).
How does Xerpa help with AI governance?
Xerpa users have no visibility of the LLM model that is actually being used. Xerpa administrators decide which AI models are configured and in which scenarios each one is used. This means a provider can be changed transparently for users, without them losing any information from their conversation history.
What security measures does Xerpa offer?
Xerpa applies security at three levels:
- Access: Xerpa has its own user and profile security system and integrates with the organisation's active directory.
- Integrations: when Xerpa connects to a file system or a corporate tool, it does so using the permissions that the user already has within that system. No user can access information they are not authorised to access.
- Prompt: administrators configure the limitations and instructions required in line with company policies.
What can I do with Xerpa's agents?
Xerpa's agents can autonomously carry out various tasks, such as running a search, generating a script, or connecting to a corporate system, among others. All you have to do is tell Xerpa what you want, and its agent orchestrator will determine which agents to use to complete the task.
Can Xerpa modify data without my permission?
No. Before creating, editing, sending or deleting any data, Xerpa shows an exact preview of the action and waits for user confirmation. This principle applies to all agents.
What data formats does Xerpa process?
Xerpa processes documents (PDF, Word, Excel, PowerPoint), images, videos, audio, and code repositories, and extracts information from multiple corporate tools: Confluence, Jira, ERPs, CRMs, and more.
Is Xerpa a product developed by Xeridia?
Yes. Xerpa is a platform developed entirely by Xeridia. Xeridia has more than 23 years of experience building complex bespoke software for large companies and holds ENS and ISO 27001 certifications in information security.
What does Xerpa offer if I'm already using ChatGPT, Claude or Gemini?
If you're already using one of these providers, Xerpa lets you keep using them, while making it easy to switch to a different provider in the future without losing information, and protecting the entire investment you make in integrations.
Want to see Xerpa with your own data, in your own environment?
If you have specific requirements around data sovereignty, AI governance and security, a demo is the best place to start.
