1 · Customer asks
Harbour Hotels SupportOnline · replies instantly
Type a message…
Powered by CevanosYour rules.Your AI agents.
AI customer service agents that answer, act and decide.
Cevanos is an AI agent for customer service that answers questions, looks things up in your systems and completes routine requests across chat, email, WhatsApp and voice. You decide what it handles alone and when it hands over to your team.
Start with answers. Enable additional actions when you’re ready.
- Human handoffRequests that need a person reach your team with the context
- Approval controlsActions above your limits wait for someone to sign off
- Decision recordsThe policy, the evidence and the outcome, on every decision
Resolution rate
Resolve 90%+ of customer queries with Cevanos.
Cevanos answers questions, takes action in your systems and decides by your policies, so most conversations close in the chat. The rest reach your team with the whole conversation in view.
Every 100 conversationsLive
90+Resolved in the chat by Cevanos, with no ticket and no wait
<10Go to your team, with everything said so far
- Answersfrom your knowledge50
- Actionsin your systems28
- Decisionsby your policies13
- Your teamwith full context9
Live conversationsResolved as they arrive
Channels, integrations and systems
AI customer support across your channels and business systems.
Connect your customer conversations with the helpdesk, CRM and business systems your team already uses. Conversational AI for customer service only helps if it can reach the record behind the question.
AI customer service by industry
AI customer service for your industry.
Handle the requests your customers bring, using your business knowledge, policies and connected systems.
Each workflow combines the information, actions and approvals needed for that request.
Controls
You set the rules. Cevanos works within them.
Set permissions, financial limits and approval requirements. Review the record behind each decision and keep your team involved where it matters.
Getting started
Connect, test and launch your AI support agent at your own pace.
Connect your knowledge, pick the actions and cards your agent can use, then switch on governance. Add policies, simulate them and plan the release. The AI understands what customers ask. Your rules decide what it may do.
Stage 01 / 05Connect
Knowledge
WebsitePages and FAQs
SyncedHelp centerArticles and macros
SyncedDocumentsPolicies and guides
SyncedChannels & systems
Chat, email, WhatsAppWhere customers reach you
LiveHelpdesk & CRMTickets and customer records
ConnectedStore & billingOrders, refunds, plans
ConnectedLive, every action passes a policy gate.
The AI reads what the customer wants and picks the policy for that intent. From there a rules engine walks your decision tree on facts pulled from your systems. No model sits in that loop, so no prompt, jailbreak or hallucination can talk its way past a rule.
AIUnderstands and talksReads the message, picks the intent, writes the reply. It never decides what is allowed.
Rules engineDecides, every time the sameWalks your policy tree on verified facts and limits. Same facts in, same outcome out.
Try a policyTry to break it
Decision record0 decisions
Every decision lands here with the policy, version and exact path that made it.
The AI can’t skip or rewrite a rule. It only picks the intent and writes the words.Claims in a message are not facts. “I’m the manager” changes nothing.Limits count across everyone. Per customer, per order and per day, in one ledger.Every decision is recorded with the policy version and the path it took.
Questions
Questions about AI customer service agents
An AI customer service agent answers customer questions and helps finish support tasks using your business knowledge and your connected tools. Cevanos adds controls on top: what information it can reach, which actions it may take, and the point at which it should bring in your team.
It means the agent works inside permissions, policies and approval limits that you set, rather than deciding for itself how far it can go. Your team controls how much authority it has and can read the record behind any decision. You can widen that authority over time as the agent earns it.
Yes. You can begin with answers from your approved content and enable connected actions separately as you review your workflows and requirements.
Yes. Requests that need approval, fall outside configured permissions or require human judgment can be handed to your team with the relevant context.
Create an account, add your knowledge sources and configure the capabilities you need. Test the agent before making it available to customers or enabling live actions.
Start with answers. Add more automation when you're ready.
Build your first agent, then explore the controls and testing tools as your needs grow.