Your systems
Approved sources.
Selected data and actions.
SECURITY & INDEPENDENCE
Keep your business foundation private. Choose who can use it, what they can do and which AI environments may receive your information.
CONTROL AT EVERY HANDOVER
Your private Atlas
Identity, permissions,
policies and monitoring.
Your chosen AI
An account and model
that fit the use case.
Atlas and the AI provider are separate environments. Any information sent to an external provider also falls under that provider’s terms, settings and processing arrangements.
PRIVATE BY DESIGN
Atlas is deployed in a dedicated private environment for your business. Your company’s knowledge, policies and integration setup are managed independently of the LLM.
Two-factor authentication is the standard for human access. Service and agent access needs its own authenticated identity, limited permissions and revocable credentials.
We agree hosting location, retention, operational access and data-processing responsibilities before connecting sensitive information.
SECURITY IS AN OPERATING PRACTICE
Identify every protected request
Authenticate people and services. Give each identity the minimum access required for the task.
Limit what an agent can do
Restrict tools, records and actions. Require approval for sensitive changes and reject requests outside the agreed scope.
See how access is used
Log relevant activity, review unusual patterns and define how alerts are handled. A connection needs oversight after launch.
Contain and respond
Revoke access, rotate credentials and isolate a connection when needed. Agree the incident process and responsibilities.
Maintain the connection
Track tools and versions in the registry. Test changes to connectors, instructions and models before rollout.
Control the data footprint
Select what may leave your systems, minimise unnecessary data and agree retention and deletion rules.
THE USE CASE COMES FIRST
Choose the environment for the work. Choose the appropriate access and security for each type of business information.
A protected workspace for your team.
For example: colleagues asking questions about margin, operations or company policy.
Illustrative setup- AI environment
- Company-managed AI workspace
- Access boundary
- Named employees, two-factor authentication and department permissions.
- What we assess
- Select suitable provider terms and verify training, retention and sharing settings for the account.
Expertise your customers can use.
For example: a Customer Plugin offering energy advice or letting an authenticated business customer retrieve a valuation report.
Illustrative setup- AI environment
- AI environments used by your customers
- Access boundary
- Only approved customer data and tools. Authenticate customers before exposing their own records; keep unrelated internal information outside this route.
- What we assess
- Publish where your audience is, subject to the platform’s supported integrations and review process.
More control over model hosting.
For workloads that need a more tightly controlled processing environment.
Illustrative setup- AI environment
- A private or locally hosted model
- Access boundary
- Choose a compatible model and an infrastructure design that meet the agreed isolation requirements.
- What we assess
- Assess performance, operations and cost alongside security. Local hosting is assessed alongside your security and compliance requirements.
INDEPENDENCE
Your company-managed AI accounts belong to your business. Customers use their own AI environments, with access to your services governed by the Customer Plugin. Atlas connects these environments to your business. Your company knowledge remains independent of the selected AI model.
Independence also includes us. Agree how your data, approved knowledge and configuration can be exported, which formats are available and how a handover works. Licensing and platform-specific components are made explicit.
That gives you a concrete path to change providers, change the setup or stop the engagement.
Bring the use case and the requirements. We’ll work through the choices together.
Contact usGOOD QUESTIONS
Is two-factor authentication standard?
Yes, two-factor authentication is the standard for human access to Atlas. Automated services and agents use dedicated credentials and scoped permissions.
How is data shared with external AI providers?
If your selected workflow sends information to an external AI provider, that transfer must be assessed separately. We define which data may be sent and review the provider account, terms and settings.
Can we choose a private or locally hosted model?
Yes, this can be considered where the model and deployment are technically suitable. We assess infrastructure, performance, operations and security requirements together.
Does Atlas guarantee regulatory compliance?
Compliance is assessed across the complete business process and technical setup, with the organisation’s security and compliance owners. Any certification or regulatory claim needs specific supporting evidence.
Who owns the account with the AI provider?
Your business owns its company-managed AI accounts for Business Plugins. Customers use their own AI accounts for Customer Plugins and authenticate separately where access to protected business services is needed.
Can we move away from Ailtitude?
Yes. We agree the export and handover of your data, knowledge and relevant configuration, including formats, responsibilities and any licensing limits. Independence is supported by an agreed export and handover process.
What stops an unknown agent from extracting all our data?
Protected tools require authenticated, authorised access with limits on what may be requested. Monitoring, usage controls and access revocation add further safeguards. Secure integrations combine authentication, permissions, monitoring and tested access controls.