Platform & Technology
Permissions do not fall between vendors when Sheba builds the agent and the knowledge layer, and chooses or fine-tunes the model for each task.
Sheba Enterprise Platform is the product. It is production-ready, and an evaluation comes before any access. Agents, enterprise knowledge and models are capability layers, which also do defined jobs inside Sheba’s products.
One product, three capability layers
Open a part to read what it does and where it runs.
- Product
Sheba Enterprise Platform
Roles are set and each task is routed to its model in one place: the operational AI layer an organisation’s teams work in.
- Capability layer
AI agents
An agent stops for a person’s approval where one is required, and otherwise acts only through the tools it was given.
- Capability layer
Enterprise knowledge
Each person’s permission sets the edge; inside it, staff get answers from the organisation’s documents.
- Capability layer
AI models
Models fitted to the use case they serve inside Sheba’s systems, built and fine-tuned by Sheba.
Where the parts meet
In Sheba’s products, ChatSheba among them, these capabilities do one defined job. In the platform, permissions are set and each task is given the model that handles it. An agent needs both of the others: the organisation’s knowledge, so it takes the right step, and a model that understands the request.
Custom systems share Sheba’s technology, standards and implementation discipline, and each is engineered for one organisation. They are not built on the platform.
Security and data boundaries
How data is handled in each deployment is set out on Sheba’s security page, and this page adds no security claim of its own.
Read how Sheba approaches securityWorking out which part fits is our job, not yours.
Which part fits comes second. First, tell us what should change in your work.
Discuss your use case