FAQ
Operations questions
Direct answers to the questions operations and customer experience leaders ask about their team, their customers and day-to-day control.
On this page
- Your team and your customers
- What does this mean for our team?
- Will our customers want to talk to an AI agent?
- Our customers expect personal service. How does this fit?
- Could it damage our brand?
- Will customers know they are talking to AI?
- What about older customers, or customers who are not comfortable with technology?
- Which languages and dialects does it handle?
- When a conversation needs your team
- What happens when the agent gets stuck?
- How does a handover work?
- What happens during a major delay, a storm or a seasonal peak?
- Can we start with the agent drafting and our team approving?
- Quality and control
- How do we check the quality of what it says?
- How do we change what the agent says?
- Who decides what the agent is allowed to do?
- Related
These are the questions operations, contact centre and customer experience leaders ask most often, answered directly. They cover what changes for your team and your customers, what happens when a conversation needs a person, and how you stay in control of what the agent says. Each answer links to the page with the full detail.
Your team and your customers
What does this mean for our team?
Your team handles the conversations that need a person: the complicated case, the upset customer, the decision that needs judgement or authority. The agent takes the routine conversations your team answers the same way many times a week, and prepares the complicated ones so your colleague starts with the facts and the history. Everything that happens is logged properly, including the routine volume, which few teams have time to do today.
See What we do and what your team does.
Will our customers want to talk to an AI agent?
What customers want is an answer, and customers who want a person get one whenever they ask. The agent tells every customer it is an AI agent, answers at any hour in the customer's language, and hands over with the full history when a person is needed. Someone with a question at eleven at night usually cares more about getting it resolved than about who resolved it.
Our customers expect personal service. How does this fit?
They get an accurate, personal answer at any hour, because the agent knows who they are and what they were told last time. When they want a person, that person picks up with the whole conversation in front of them. You can also set particular customers or accounts to go straight to a named person or team, every time.
See Escalation rules.
Could it damage our brand?
The agent speaks in the tone you set. Tone is configured with you as part of the build: how formal, warm or brief the agent is, and how it handles a complaint, in each language. You hear the agent, including its voice, before it goes near a customer, and topics you place out of scope are never discussed.
Will customers know they are talking to AI?
Yes. The agent identifies itself as an AI agent and offers a person whenever a customer asks for one. A customer who knows what they are talking to, and gets their answer, is far better placed than one who later feels misled.
What about older customers, or customers who are not comfortable with technology?
They call your usual number and say what they want, in their own words. There is no app to install and no menu to work through, and the agent asks a question back when something is unclear. Anyone who would rather speak to a person can ask for one at any point.
See Voice: inbound and outbound calls.
Which languages and dialects does it handle?
Arabic in its Gulf, Egyptian, Levantine and Modern Standard forms, English, and many other languages, including Urdu, Russian, Chinese and Spanish, and switching between Arabic and English within a sentence. The agent works out which language a customer is using from their first few words, replies in it, and remembers the customer's preference for next time. Languages beyond Arabic and English are configured during implementation where your customers need them.
See Supported languages and Arabic dialects.
When a conversation needs your team
What happens when the agent gets stuck?
It hands the conversation to your team, so the customer is never left without a way forward. Handover combines rules your team sets for each type of question with a confidence threshold measured for the model before deployment, so a consequential answer the agent is not sure of goes to a person. You set the triggers: a topic it should not handle, a customer getting frustrated, a request it cannot complete, or an account you always want a person on.
See Escalation rules.
How does a handover work?
The person picking up receives the full conversation and a written brief: who the customer is, what they asked, what has been established and what is outstanding. The brief arrives in the system your team works in, whether that is your contact centre platform, your CRM (the system that holds your customer records) or the Telonic console, so the customer does not repeat themselves. Where you choose, the agent picks the conversation back up once your colleague is done; this is configured during implementation.
What happens during a major delay, a storm or a seasonal peak?
The agent holds many conversations at the same time, so capacity follows demand rather than the rota. It takes the routine volume, such as rebooking questions after a cancelled flight or claim status after a storm, so your team is free for the people who need someone. Overflow routing, configured during implementation, sends calls to the agent when your queue passes a threshold you set.
See Voice: transfers, callbacks and overflow and Travel and hospitality.
Can we start with the agent drafting and our team approving?
Yes. For email, the agent can prepare each reply and hold it for a person on your team to approve before it is sent, which suits a sensitive inbox. Actions you mark as sensitive in your systems can also wait for approval. Both are configured during implementation.
See Email: drafts for human review.
Quality and control
How do we check the quality of what it says?
Every conversation is scored against your own quality standard, not a sample. Your team can search every conversation by keyword and by meaning, open any one to see what was looked up and why, and follow outcomes and reasons for contact in the console. Supervisors can also watch live conversations, which is configured during implementation.
See Quality review on every conversation.
How do we change what the agent says?
Most answers come from your own documents and systems, so updating the source updates the answer. Your team adjusts topics, limits, escalation and handover rules, approvals and retention through the controls it holds in the console. Larger changes, such as a new workflow, are made by Telonic with you, and every change is tested before it goes live, recorded, and can be reversed.
See How changes go live and The controls your team holds.
Who decides what the agent is allowed to do?
Your team does, before go-live and at any point afterwards. You decide which requests the agent completes, which it prepares for a person, and which always stay with your people, and those limits are built into what the agent is able to do. Decisions that need judgement or authority, such as a refund outside policy or a claims decision, are not available to it at all.
See Decision boundaries.