AI agents that talk to your customers, on the phone and on your site
A customer-facing AI agent answers your customers on the phone and on your website from one setup and one knowledge base. The Call Center Voice Agent picks up the phone line, the Chat AI Agent answers the site, and both capture the lead, qualify it against your criteria, book the meeting, and hand your team a conversation with full context.
What Your Agents Handle
- Answer the phone on the first ring
- Reply to website chats in seconds
- Ask your qualifying questions every time
- Book meetings straight into your calendar
- Hand off to a person with the full transcript
- Call new leads back before they cool off
What is a customer-facing AI agent?
A customer-facing AI agent is software that holds real conversations with your customers by phone or chat, using your knowledge base and instructions rather than scripts. On the Tridan AI platform it answers questions, qualifies leads, books meetings in your calendar, works inside tools such as Gmail, Jira, and Zendesk, and hands over to a person when needed.
See the AI Call CenterWhich channels does one agent cover?
One agent setup covers two channels. The Call Center Voice Agent answers inbound calls and runs outbound campaigns through your phone system, while the Chat AI Agent answers visitors on your website and on WhatsApp. Both use the same knowledge base, the same qualifying questions, and the same handoff rules, so customers get consistent answers wherever they reach you.
Your phone line
- Answers inbound calls on the first ring, many at once
- Runs outbound follow-up and reminder campaigns
- Natural voices from ElevenLabs, Deepgram, and OpenAI
- Telephony over Twilio with LiveKit real-time audio
Your website chat
- Answers in the context of the page the visitor is reading
- Type or talk, voice mode lives in the same widget
- Pre-chat forms and in-conversation lead capture
- One embed snippet, no development work
- CustomerWebsite chat, WhatsApp or phone
- One agentSame knowledge base and rules
- ToolsCalendar, tickets, order lookups
- AnswerResolved in the channel they chose
- Handoff inboxYour team takes the rest
Which jobs can the agent be configured for?
The same agent is configured for the job you need covered: a reception agent that answers and routes, inbound sales and lead qualification agents that ask your questions and book meetings, a support agent that resolves questions from your documentation, an outbound caller that follows up on leads, or an internal agent for your own team. Each role gets its own instructions, tools, and handoff rules.
Reception & Front Desk
Answers the main line and the site chat, routes the caller to the right team, and books the appointment instead of taking a message.
- Call routing
- Appointment booking
- After-hours cover
Inbound Sales
Handles pricing and product questions the moment they are asked, qualifies against your criteria, and books the meeting while the interest is still there.
- Pricing questions
- Meeting booking
- Instant response
Lead Qualification
Asks your qualifying questions on every inbound conversation and passes your team only the leads that actually fit.
- Custom qualifying questions
- Structured lead records
- Pipeline sync
Customer Support
Resolves the repetitive questions from your own documentation and escalates everything else with a full transcript attached.
- Answers from your docs
- Ticket creation
- Escalation rules
Outbound Follow-Up
Calls new leads back, chases open quotes, confirms appointments, and runs reminder campaigns from your list.
- Lead callbacks
- Reminder campaigns
- Outcome logging
Internal Knowledge Assistant
Answers your own team's questions about policy, product, and process from the same knowledge base your customers reach.
- Shared knowledge base
- Role-based access
- Multi-tenant workspaces
Your Own Agent
Most teams need something between two of the roles above. We configure the agent around your actual workflow, your intake questions, your booking rules, your escalation policy, and stay with you until it is handling live traffic.
- Configured with you, not by you
- Trained on your documents and your website
- Connected to your calendar, inbox, and ticketing
- CustomerAsks a question or wants to book
- One knowledge baseYour documents and website
- Support agentAnswers and resolves
- Sales agentQualifies and quotes
- Booking agentBooks and reschedules
- Handoff inboxYour team finishes the rest
Every Agent Ships With This
Not add-ons, this is what an agent is on this platform.
Your Knowledge, Not the Internet's
Upload documents, crawl your website, or sync Confluence, Notion, and Zendesk. Answers are grounded in your content.
Human-Handoff Inbox
When a conversation needs a person, it lands in a shared inbox with the full transcript. The customer never repeats themselves.
Lead Capture & Qualification
Pre-chat forms and in-conversation capture collect the contact details and qualify against your criteria before the lead reaches your team.
Calendar and Inbox Actions
Connected to Gmail, Google Calendar, Outlook, Jira, Zendesk, and WhatsApp, so the agent books and files instead of only replying.
4 Leading AI Models
OpenAI, Anthropic, Meta, and DeepSeek. Choose the model per agent and change your mind later.
Supervisor Orchestration
A supervisor agent can route one request across specialist agents and combine their answers into a single reply.
Agents That Bring You the Problem Before a Customer Does
Answering well is the baseline. The next step is an agent that reads what your business already produces and tells your team what needs attention, the quote nobody chased, the caller who has phoned three times this week, the intake question that keeps losing people.
Reads What You Already Have
CRM activity, email, support tickets, call outcomes and campaign events, the systems you run today, with nothing new for your team to fill in.
Brings You What Changed
The quote nobody chased. The customer who has called three times this week. A quiet week gets a quiet report.
Shows Its Evidence
Every finding links back to the call, ticket or record it came from, so your team can check it in one click.
- Your dataCalls, chats, quotes, tickets
- Proactive agentReads and compares daily
- FindingsWhat needs attention, with evidence
- Your teamDecides and acts
- CustomersHear back before they chase you
Proactive agents are in early access. We onboard a limited number of companies at a time so our team can configure each one against your data sources and your definition of what is worth flagging, most go live in two to six weeks. It is not a feature you switch on in the dashboard.
Why Teams Put an Agent in Front of Customers
People asking questions are people ready to buy. The only variable is whether anyone answers.
The First Real Answer Wins
People contacting you are contacting your competitors too. Whoever holds an actual conversation first usually keeps the deal.
Demand Doesn't Keep Office Hours
Calls and chats arrive at night, on weekends, and mid-holiday. The agent handles every one of them from your knowledge.
An Unanswered Call Is a Lost Lead
A conversation that gets answered can end in a booking. One that doesn't ends on someone else's phone line.
Your Team Keeps the Real Work
The agent absorbs the repetitive volume; escalations reach your people with full context in the handoff inbox.
Common questions about customer-facing AI agents
How does the agent know about our business?
The agent works from a knowledge base built from your documents, website, price lists, and connected tools such as Confluence, Jira, or Notion, plus written instructions that define its role, tone, and limits. It looks up answers at the moment a question is asked, so updating a document updates the agent. The language model itself is never trained on your data.
What happens when the agent cannot answer?
It says so and hands the conversation to a person according to the rules you set: the voice agent transfers the call to the operator number you define, and the chat agent places the conversation in the handoff inbox with the full transcript. Your team continues from there without the customer repeating anything, and those cases feed the next round of improvements.
Can the agent book meetings and update our systems?
Yes. Through the ready-made integrations the agent books meetings in Google Calendar or Outlook, sends confirmations through Gmail, Outlook, or SMTP, creates tickets in Zendesk or Jira, and messages customers on WhatsApp. Custom API tools and MCP servers connect your own systems. Each agent only gets the actions you enable, and every action is visible in the conversation.
Which languages does the agent speak?
The agent answers in the languages your customers use. The chat and voice agents are built on large language models that handle many languages, including English, Serbian, German, and Spanish, and the voice agent uses voices from ElevenLabs, Deepgram, and OpenAI, so you can choose a natural voice for each language. Instructions and knowledge can be written in one language and answered in another.
Which AI models power the agent?
The platform uses four leading model families: OpenAI, Anthropic, Meta, and DeepSeek, selectable per agent, so a routing agent can use a fast model while a support agent uses a stronger one. In self-hosted deployments, open models can also run locally through Ollama. Voice agents add speech from ElevenLabs, Deepgram, or OpenAI and telephony through LiveKit and Twilio.
How do we try it before committing?
You can talk to a live demo agent on the agent widget page right now, by chat or by voice, and see how it answers, qualifies, and books. For your own use case, book a demo: we walk through how enquiries reach you today, show the agent handling a call and a chat on your scenario, and tell you honestly what fits and what does not.
Last updated · Reviewed by Nemanja Milivojevic, Founder
See It Answer Your Calls
Book a 30-minute demo. We will walk through how enquiries reach you today, show the agent taking a call and a chat, and tell you honestly what fits and what doesn't.