What is PRESHos Platform
PRESHos Platform is the governed AI operating layer for running complex agent workloads on real application context. Organizations use it to put agents to work on production systems (tickets, devices, users, configs, documents), with isolation, approvals, and audit enforced from the first action.
PRESHai (the company) implements and operates the platform for customers, primarily in the IT channel. Customers get governed outcomes; PRESHai carries the implementation and operational load.
The problem it solves
Section titled “The problem it solves”Agent demos are easy. Production agent work is hard because real work means:
- Real context: live applications, not sample data. Agents must read and act on the systems the business actually runs on.
- Real blast radius: agents that can change things can break things. Every write needs a boundary.
- Real accountability: regulated IT work needs to answer who approved what, when, and on which system.
PRESHos exists so an organization can say yes to agents without accepting uncontrolled side effects: every session runs isolated, sensitive actions pause for approval, and everything is recorded.
What PRESHos is
Section titled “What PRESHos is”- An operating layer, not a model. It sits between agents and your stack, governing what agents can see, touch, and change.
- Workload-oriented. It is built for sustained, multi-step agent work (investigations, remediation, provisioning, reconciliation), not one-shot Q&A.
- Governed from the first action. Isolation, identity, approvals, and audit are part of the session model itself, not added later.
- Operated with you. PRESHai implements and runs the platform for the customer (AI Implementation and Managed AI Ops services), rather than handing over a toolkit and leaving.
What PRESHos is not
Section titled “What PRESHos is not”| Claim | Reality |
|---|---|
| A consumer chatbot | No. PRESHos runs operational agent workloads for organizations, on their systems, under governance. |
| A PRESH-internal-only tool | No. It is a platform customers operate on. Customer admins, operators, and builders all have defined roles. |
| A connector marketplace where everything works out of the box | No. Connectors are provisioned deliberately. Code existing for a connector is not the same as that connector being live and connected for a tenant. |
| Public self-serve SaaS | No. There is no open signup. Tenants are provisioned by invitation, then onboarded through a guided flow (see Day-0 for customer admins). |
The three guarantees
Section titled “The three guarantees”Every governed agent session on PRESHos holds to three guarantees:
- Isolation: agent work runs inside an Agent Environment, scoped to exactly the context and tools it needs. Sessions cannot wander into systems they were not granted.
- Approvals: actions above a risk threshold pause for a human decision. Approvers see what the agent wants to do, on what, and why, then allow or deny.
- Audit: sessions, tool calls, approvals, and outcomes are recorded so the organization can reconstruct what happened and prove it.
If a workload cannot promise all three, it does not run as a governed session.
What first value looks like
Section titled “What first value looks like”The first value a new tenant sees is deliberately small and concrete: sign in, open an environment, run one governed agent session. An agent does real work, pauses for approval where it should, and leaves a complete audit trail afterward. Everything else (more environments, more connectors, more policy) scales that loop up.
See where that work happens in Agent Ops, and where tenant setup lives in Settings.
Where to go next
Section titled “Where to go next”- Core concepts: tenant, hostname, org vs. company, environments, sessions, approvals, connectors.
- Day-0 for customer admins: from invite to first governed session.
- Agent Ops: the home console, sessions, and agent config where daily work runs.
- Settings: org, people, security, integrations, and MCP servers.
- Naming guide: PRESHai vs. PRESHos Platform vs. services vs. Agent Ops.
- Learning paths: role-based guided routes through this material.