The Agent manages the process but knows when to hand the conversation to a human
Autonomy does not mean having no boundaries. The Agent uses approved scenarios, shared knowledge and permissions, while every decision remains auditable.
Late arrival and a non-standard request
The guest first asks on WhatsApp about entering the hotel at night, then requests instructions by email and, during a phone call, asks for an exception to the standard key collection process.
Automated handling and controlled escalation
The Agent answers routine questions, maintains the shared context and routes the exception to the right person without promising the guest an outcome.
- one history across all channels
- automated actions only within the permitted scope
- immediate handover with a summary
Coordinating routine communication, not replacing the entire reception team
The Agent manages cases that can be handled using reliable knowledge, clear rules and available permissions. Unusual needs, new information, decisions and sensitive situations remain with a human.
Coordinates channels
It maintains one case across chat, messaging platforms, email and telephony, keeping track of agreed details.
Follows routine scenarios
It answers independently and carries out the next steps only when the approved conditions are met.
Escalates with full context
It recognises the limits of its permissions and passes the case to the right person with its history, risks and required action.
How does the Communication Agent manage a case?
Identifying the guest and their need
The Agent identifies the topic, channel, language and existing context within its available permissions.
Selecting an approved scenario
It checks the knowledge, conditions, risks and possible actions before starting automated handling.
Handling and monitoring
It manages the conversation, maintains continuity and monitors quality, gaps and changes in sentiment.
Closing or handing over
It closes a routine case with the outcome recorded or escalates it to a human with a full summary.
The Agent works from shared knowledge, history and explicit permissions
It is based on approved sources, active email and voice scenarios, conversation continuity and a panel for assessing outcomes and exceptions.
What does AI use?
- text and voice conversations
- approved knowledge and scenarios
- permissions, boundaries and escalation rules
- case continuity, quality and risk signals
What does the team receive?
- a routine case handled across one or more channels
- a history of responses, decisions and sources used
- escalation to the right person with full context
The Agent has a clear scope of autonomy and an immediate route to a human
The hotel approves every scenario, channel and action. The panel shows outcomes, quality, gaps and escalations, while the team can narrow or disable the scope at any time.
Staff set the policies and permissions, handle exceptions, review audits and approve every extension of autonomy.
The Agent does not make new business decisions, invent facts, replace human empathy in sensitive situations, act beyond its permissions or hide escalations.
Routine conversations handled consistently, exceptions managed thoughtfully
The guest receives consistent information and next steps regardless of the contact channel.
The Agent manages repetitive conversations, handovers and statuses without constant staff involvement.
The boundaries, sources and decisions are visible, and cases requiring human input are passed on immediately.
The Agent combines automation with a control layer
The complete process uses email and telephony automation, the Conversation Panel, shared knowledge and text channels connected to the Central Communication Panel.
Related process elements
Frequently asked questions
Does the Communication Agent replace reception?
No. It takes over routine, clearly defined conversations, while decisions, exceptions, sensitive situations and operational actions are passed to people.
Does it work across all channels from day one?
It does not have to. The scope can be introduced in stages, starting with one channel and a few low-risk cases.
How do we know what the Agent has done?
Every conversation, source used, decision, action and escalation is recorded in the central panel.
When does the Agent escalate a case?
When reliable knowledge is lacking, a decision or empathy is needed, a risk arises or the requested action exceeds its permissions.