AI Security & Control

Give AI capability.
Not unrestricted access.

At Ongo, we believe AI should strengthen how organisations work without losing control over who can act, what they can access and which services they can use. That principle is built into Agility.

Agility is a live, self-hostable operating environment designed to connect work, systems and AI within defined identity, permission and deployment boundaries.

Five-layer stack: AI & Agents, Workflows & Execution, Context & Knowledge, Permissions & Identity, Your Systems & Data

Our approach

Control is part of execution.

AI creates value when it can work. It remains trustworthy when it works within clear boundaries.

Agility connects work to identity, permissions and existing user authority, so AI operates inside your environment rather than outside it.

Every action is permissioned, attributable and controllable — ensuring AI can move work forward while your organisation retains control.

Deployment routes: on-premises, private cloud, customer data centre and supported hosting

Deployment & data

Your environment.*
Your control.

The Agility operating environment does not require a public cloud. You can run it in the environment that fits your requirements, and external services are optional.

(*) Talk to Ongo for full hard- and software requirements.

Confirmed routes

  • On-premises
  • Private cloud / customer-controlled data centre
  • Supported hosting
  • Local-model connection option

The control model

From identity to outcome.

Every action in Agility is permissioned, controllable and attributable. Control is not a separate layer — it is built into how work gets done.

Current AI participation inherits existing user authority and can be bounded by customer-approved scope. Separate agent/service identities, action-specific scopes and policy-based approvals are still being developed.

  • Identity

    Who or what is acting?

  • Permissions

    What is it allowed to do?

  • Boundaries

    Which data, systems and services may it access?

  • Evidence

    What remains attached to the work?

Data & model control

Your data should go where you
decide it can go.

Agility does not require a public cloud deployment. Organisations decide which local models, approved external models and selected services are intentionally connected, and what data and context those routes may receive.

Exact provider, region, residency and transfer choices remain deployment-specific.

Your organisation

Data
Systems
People
Policies

Agility

Identity & access
Context & permissions
Tools & actions
Work & evidence

Models & services

Local model (on-premises)
Approved external model
Selected services
Cream and teal blocks representing the Vault knowledge layer

Vault

Turn organisational knowledge into
operational advantage.

Vault extends Agility with organisational knowledge while keeping that knowledge inside the access and permission context around the user and the work. Permission-aware retrieval makes relevant organisational context usable for people and AI without treating the entire knowledge base as unrestricted model access.

Curved architectural wall with glass and greenery

Responsibility

Technical controls.
Clear responsibility.

Agility provides technical controls for identity, access, AI participation and connected services. Organisations remain responsible for defining who may access their data, which models and services may be used, and what authority is granted.

Scale AI with control.