Responsible AI · 2 August 2026
You should know when AI is helping.
AT AI produces machine-generated suggestions from client-approved instructions and sources. It is a support tool, not a human or an independent decision-maker.
What the system does
Depending on the enabled service, AI may answer questions, retrieve approved knowledge, summarise conversations, classify requests or suggest follow-up actions. The active client, service and provider settings determine what is available.
Important limitations
- Output may be inaccurate, incomplete, outdated or inappropriate.
- Source grounding reduces error but does not guarantee correctness.
- Users should verify important information with the client or a qualified person.
- AT AI must not make autonomous employment, credit, insurance, medical, legal or similarly consequential decisions.
Human control
Clients approve knowledge and configuration, test the service before activation, monitor results and provide a human escalation path. Write actions and optional channels stay off unless specifically reviewed and enabled.
Data and providers
Conversation content may be sent to the configured AI provider to generate a response. OpenAI is not enabled for the live pilot until an API key and route are approved. Provider use is governed by the client agreement, privacy notice and data-processing terms.
Report a concern
Do not include unnecessary sensitive personal data. Report harmful, incorrect or unexpected output to the client operating the assistant or email sureshkardam@gmail.com.