Volerai / AI assistants
Create assistants that know your documents and tools
Use Volerai assistants to work with documents and data in context. Each agent can use a different model, and local models can be connected within the installation. Tools and permissions are defined for the scenario. Detailed logging, role-based access, and who may use each agent can be set.
Logging, roles, and agent access
Actions are written to a detailed log. Roles separate viewing from running. You can define who may use each agent. Human approval can be part of the check step in the scenario. Writing data can stay locked until the scenario allows it.
Multiple agents
More than one assistant can work in the same scenario. Each can use a different model, including a local model. Suggestions are reviewed. Writing or changing data stays locked until the scenario allows it.
Explain the information
Help reviewers see where a value came from in the document. Handle questions about the document inside the working context.
Evaluate suggestions
Create correction suggestions for missing or inconsistent fields. Separate reviewing a suggestion from applying it, according to the process.
Attach a model to the agent
Each agent can run with a different model. Local models are also an option within the installation, and the model step can run on your infrastructure. The agent can carry out the operations defined in the scenario. A suggestion is not applied on its own without the permission you define.
Try it before you publish
Evaluate the assistant definition with sample questions and documents. Follow which tools were used from the run records.