AI that answers, analyzes, and automates built for your business
Tridan AI builds and runs AI agents for business: a Chat AI Agent for your website and messaging channels, and an AI Call Center that answers and places phone calls. Both run on OpenAI, Anthropic, Meta or DeepSeek models, connect to tools like Gmail, Jira and Zendesk, and deploy on our cloud or inside your own infrastructure.
Or call our AI agent now:+381 11 635 5917
- Voice & chat agents
- Call analytics
- Integrations & automation
- Our cloud or your infrastructure
What does Tridan AI do?
Tridan AI delivers six things: an AI Call Center for inbound and outbound phone calls, a Chat AI Agent for your website, voice analytics for every recorded call, integrations that automate routine work, custom AI development, and consulting with self-hosting on your own servers. Pick one solution or combine them; we build, integrate, and support everything we ship.
AI Call Center
Voice agents that answer inbound calls and run outbound campaigns, 24/7.
Explore the call center
Chat AI Agent
A website chat agent that answers, qualifies, and books, in your customers' language.
Explore the chat agent
Voice Analytics
Every call transcribed and analyzed: outcomes, sentiment, and agent coaching.
Explore voice analytics
Integrations & Automation
Agents connected to your calendar, email, helpdesk, and CRM to handle routine work.
See integrations
Custom AI Development
Agents and workflows built around your processes, from first audit to launch.
See our services
Consulting & Self-Hosting
AI strategy, and the full platform installed on your own servers.
Explore self-hostingHow does the AI Call Center answer calls?
The Call Center Voice Agent picks up on your phone line through Twilio and LiveKit telephony, understands what the caller wants, answers from your knowledge base, books appointments or creates tickets, and hands the call to a person with full context when needed. The same agent places outbound calls for reminders, lead qualification, and surveys.
Inbound: every call picked up
- Answers every call instantly, no hold music
- Understands intent and resolves or routes the call
- Books appointments and creates tickets on the spot
- Escalates to a human with full context
Outbound: campaigns that run themselves
- Calls whole contact lists from your CRM or CSV
- Qualifies leads and schedules meetings automatically
- Reminders, follow-ups, surveys, and win-backs
- Every call logged with transcripts and outcomes
Telephony by Twilio & LiveKit · Voices by ElevenLabs, Deepgram & OpenAI
What can the Chat AI Agent do on a website?
The Chat AI Agent is a website widget that reads the page your visitor is on, answers from your documents and website in the visitor's language, captures contact details as a qualified lead, and hands the conversation to a person through the handoff inbox when needed. It runs on the AI model you choose and can call your own tools.
Page-aware website widget
The embedded widget reads the page your visitor is on and answers in context: product pages, docs, or pricing.
Multi-model intelligence
Run each agent on OpenAI, Anthropic, Meta, or DeepSeek models and change your mind anytime.
Connect your own tools
Connect internal systems and APIs so agents can take real actions such as booking, updating, and fetching, not just answer questions.
Teams & organizations
Separate workspaces for every team or brand, with roles and budget limits to keep them cleanly isolated.
Chat agent or voice agent: which one do I need?
Choose the Chat AI Agent when customers reach you through your website or messaging apps and you want instant written answers and lead capture. Choose the Call Center Voice Agent when the phone is your main channel and calls must be answered or placed at any hour. Many teams run both on the same knowledge base.
Swipe sideways to see the full table
| Chat AI Agent | Call Center Voice Agent | |
|---|---|---|
| Channel | Website widget and messaging channels such as WhatsApp | Phone calls, inbound and outbound, via Twilio and LiveKit |
| Typical jobs | Answers product and support questions, qualifies visitors, captures leads, books meetings | Answers the line, books appointments, creates tickets, runs reminder and lead qualification campaigns |
| Human handoff | Hands the conversation to your team in the handoff inbox with the full transcript | Transfers the call to a person with the context collected so far |
| Languages | Replies in the languages you configure, including English, Serbian, German, and Spanish | Speaks in the languages and voices you configure (ElevenLabs, Deepgram, OpenAI) |
| Deployment | Tridan AI cloud or your own infrastructure | Tridan AI cloud or your own infrastructure, with your PBX or SIP trunk |
| Best for | Websites with steady visitor questions and lead forms | Businesses where the phone is the main channel: clinics, services, sales teams |
Bottom line: start with the channel where customers already contact you most; both agents share one knowledge base, so adding the second later is straightforward.
Powered by 4 leading AI model providers
- OpenAI
- Anthropic
- Meta
- DeepSeek
Pick the best model per agent and switch anytime.
How model selection worksWhat can the proactive AI analyst tell me?
Ask about sales, margins, stock, or collections, and the agent answers from your own data with read-only queries. In a pilot, it also brings you what changed: the product that is slipping, the margin leak, the late payer. Nothing in your systems changes unless you approve it.
Read-only query · approved by youAsks your data
Plain-language questions answered with real, read-only queries over your sales, accounting, or warehouse data.
You approve every step
Anything that would change or send something waits for your approval. Installed on your server, with the model you choose.
Findings in a pilot
In the pilot, the agent and our team bring you regular findings: what moved, why, and what to do about it.
It reads. Nothing changes in your systems unless you approve it.
Proactive findings are in early access, delivered in a pilot configured for your company. Start with a free 5-day analysis from one export.
Where do the agent's answers come from?
Every answer comes from your own content: uploaded documents, your website, and connected wikis such as Confluence, Notion, and Zendesk. The agent retrieves the relevant passage before it replies, so it does not answer from the open internet, and built-in PII detection keeps personal data out of model inputs.
Documents & files
PDF, Word, Excel, and CSV files, parsed, chunked, and searchable in minutes.
Websites & wikis
Crawl your site or sync Confluence, Notion, and Zendesk continuously.
Deep file retrieval
Finds the exact passage you need in seconds, even across hundreds of files.
PII protection
Built-in detection and anonymization keeps personal data out of model inputs.
Which tools does Tridan AI connect to?
Tridan AI agents connect to Gmail, Outlook, Google Calendar, Confluence, Jira, Notion, Zendesk, WhatsApp, SMTP, and LinkedIn out of the box, and to your CRM, ERP, or database through custom API tools or an MCP server. Knowledge stays in sync and actions land where your team already works.
CRM & customer data
Connect your CRM through its API, so conversations look up customers and update their records.
Email & calendars
Send emails, book meetings, and check availability without leaving the conversation.
Docs & knowledge bases
Pull answers straight from your wikis, files, and internal documentation.
Your own systems
ERP, internal tools, or databases, connected through custom API tools or an MCP server.
Where does my data live?
On Tridan AI's managed cloud, with EU server regions available on request, or entirely inside your own infrastructure: the full platform, including the AI models via Ollama, installs in your VPC or on-premises servers. In both cases your conversations, documents, and calls stay under your control and never train anyone's model.
Managed cloud
We run the infrastructure for you, with EU-only server regions available on request.
Full platform on your infrastructure
Chat, voice, knowledge, and AI models, including local models via Ollama, install inside your own VPC or on-prem servers.
Built for GDPR obligations, DPA on request
Keep full control over where data is processed and stored. Enterprise customers can request a signed Data Processing Agreement.
Security for organizations
PII anonymization, content moderation, per-tenant API keys, and separate workspaces with roles for every team.
Prefer to start on your own?
The Tridan AI platform is also available as a self-serve subscription: start on the Free Plan, then move to a paid plan with chat, voice minutes, and AI agents included, with pricing confirmed on a call.
Free Plan
Free
- 30,000 tokens monthly
- 100 chat messages
- 5 voice minutes
- 0 AI agents
Starter Plan
Call
- 600,000 tokens monthly
- 500 chat messages
- 30 voice minutes
- 1 AI agent
Business Plan
Call
- 1,000,000 tokens monthly
- 1,000 chat messages
- 100 voice minutes
- 3 AI agents
Enterprise Plan
Call
- 5,000,000 tokens monthly
- 5,000 chat messages
- 500 voice minutes
- 10 AI agents
How do we get started?
Getting started takes four steps: a short audit of your calls, chats, and processes; a design session where we agree what the agent does and which systems it connects to; a build and test phase on your real scenarios; and a launch with monitoring and ongoing improvement. Your team needs no in-house AI expertise.
Audit
We map your calls, chats, and processes and find where AI pays off first.
Design
Together we define what the solution does, which systems it connects to, and how success is measured.
Build & integrate
We build the agents and integrations and test them on your real scenarios.
Launch & support
We go live, monitor results, and keep improving the solution with you.
Frequently asked questions about Tridan AI
Short answers to the questions we hear most often. For the full list, see our FAQ page.
What is Tridan AI?
Tridan AI is a Belgrade-based company that builds and runs AI agents for business. Its platform provides a Chat AI Agent for websites and messaging channels and an AI Call Center whose voice agent answers inbound calls and places outbound calls. Both agents answer from your own documents and website, connect to tools such as Gmail, Jira, and Zendesk, and run on the Tridan AI cloud or inside your own infrastructure. The company also offers custom AI development and consulting.
Which AI models does Tridan AI use?
Tridan AI runs on models from four providers: OpenAI, Anthropic, Meta, and DeepSeek. You pick a model per agent, so a support agent can run on a fast, lower-cost model while a sales agent uses a more capable one, and you can switch at any time without rebuilding the agent. Self-hosted installations can additionally run open models locally through Ollama, so prompts and documents never leave your servers.
Can I run Tridan AI on my own servers?
Yes. The full Tridan AI platform, including chat, voice, the knowledge base, and the AI models, can be installed inside your own VPC or on on-premises servers. Local models run through Ollama, so conversations, documents, and call recordings stay within your network. Self-hosting is offered as the Hosting Plan, scoped on a call, and it suits organizations with strict data residency or compliance requirements. The managed cloud remains available for teams that prefer not to operate infrastructure.
How much does Tridan AI cost?
Tridan AI has a Free Plan you can start on today, with a monthly allowance of tokens, chat messages, and voice minutes. The paid tiers are the Starter Plan, Business Plan, and Enterprise Plan: each raises those limits and adds AI agents, and from the Business Plan up you also get premium models. The Hosting Plan covers self-hosted installations. Paid plan pricing is confirmed on a short call so it matches your call volume and channels; the pricing page compares every plan side by side.
Which tools does Tridan AI integrate with?
Out of the box, Tridan AI agents connect to Gmail, Outlook, SMTP email, Google Calendar, Confluence, Jira, Notion, Zendesk, WhatsApp, and LinkedIn. Inside a conversation an agent can send an email, book a meeting, create or update a ticket, and look up a document. Systems without a ready-made connector, such as your CRM, ERP, or internal database, are added through custom API tools or an MCP server, and your developers keep control of what each tool is allowed to do.
How long does setup take?
It depends on the channel and how many systems the agent must talk to. A website Chat AI Agent trained on your existing documents and site can usually be live within days, because the widget is a single script tag. An AI Call Center takes longer: we connect your phone number or SIP trunk, design the call flows, and test them on real scenarios before launch. The audit at the start of every project gives you a concrete timeline for your case.
Does the voice agent speak Serbian, German, and Spanish?
Yes. Both the Call Center Voice Agent and the Chat AI Agent work in the languages you configure, and Tridan AI itself operates in English, Serbian, German, and Spanish, so those four are well covered. Voices come from ElevenLabs, Deepgram, and OpenAI, and you choose the voice and language for each agent. If your customers use several languages, tell us on the call and we will recommend the configuration that fits your case.
Can I try Tridan AI before I commit?
Yes. The quickest way is the live demo page, where you can chat with a Chat AI Agent trained on real business data and see how it answers, captures a lead, and hands over to a person. You can also create a Free Plan account on the platform and try it with your own documents. If you want to see the AI Call Center, book a call and we will walk you through a live inbound and outbound scenario.
Last updated · Reviewed by Nemanja Milivojevic, Founder
Tell us what your team spends too much time on
Book a 30-minute call. We'll look at your calls, chats, and processes, and show you where AI can take over and what it would take.
Or call our AI agent now:+381 11 635 5917