Intelex develops and integrates artificial intelligence to support useful, accountable outcomes. We apply governance proportionate to the purpose, data, autonomy and potential impact of each system.
Our principles
- Clear purpose and measurable benefit
- Lawful and appropriately governed data
- Human oversight proportionate to impact
- Security, privacy and resilience by design
- Testing for quality, bias, misuse and failure
- Transparency about material limitations
- Traceability and accountability
- Ongoing monitoring and controlled change
Agentic systems
AI agents are given bounded objectives, approved tools, least-privilege permissions and defined escalation. High-impact, irreversible or externally binding actions should require proportionate human approval. Agent instructions, tool use and outcomes should be logged to the extent appropriate and lawful.
Evaluation
AI is evaluated against representative tasks and foreseeable misuse. Measures may include factual grounding, task completion, fairness, security, privacy, robustness, cost, latency, escalation and user understanding. A system should fail safely when evidence or confidence is insufficient.
Third-party AI
External models and AI services are assessed for security, data use, location, contractual protection, lifecycle risk and operational dependency. Customer or confidential data must not be entered into unapproved public AI tools.
Questions and concerns
Concerns about an Intelex AI capability may be sent to uk@intelexsystems.com with the subject ‘Responsible AI’. Reports are reviewed through appropriate product, security, privacy and risk governance.