Every channel in one place
One assistant on your website, WhatsApp, Viber, Instagram, Facebook Messenger and email, with the same tone, the same information and a single conversation history.
AI chatbot development
AI chatbots that answer like your best employee
We build AI chatbots for your website, WhatsApp, Viber, Messenger and email that answer from your knowledge base, capture enquiries and book appointments, then hand complex conversations to your team. All on our own fine-tuned models, with no data sent to outside AI services.
- Our own models tuned to your industry
- English, Montenegrin, Russian, Turkish and more
- Live in 2–6 weeks
Answered from knowledge base
Handed off to an operator
- customer replies across every channel
- 24/7
- channels: web, WhatsApp, Viber, Messenger, email
- 5
- weeks from workshop to launch
- 2–6
- starting price for a website AI chatbot
- €800
Why an AI chatbot
Customers message around the clock, but your team can’t always be online
Most of the messages a business receives repeat day after day. When a reply is slow, customers go to a competitor, and your staff spend hours explaining the same things.
Enquiries go unanswered after hours
Messages arrive in the evening, at weekends and in peak season when the team is swamped. Every hour of waiting lowers the chance an enquiry becomes a booking or a sale.
The same questions, hundreds of times
Prices, opening hours, delivery, cancellation terms and order status. Your team repeats the same answers instead of selling and handling the cases that need them.
Conversations scattered across channels
Some customers use the website, others WhatsApp, Viber or Instagram. Information gets lost, answers differ, and nobody has the full picture.
What you get
A chatbot that knows your business, not one that just chats
Every chatbot runs on our fine-tuned models and is connected to your content and systems, so it gives accurate answers and gets real tasks done.
Answers from your knowledge base
It learns from your price lists, manuals, terms and FAQs. When the information isn’t there, it says so and offers a contact instead of making things up.
Handoff with full context
Complaints, special offers and unhappy customers are recognised and passed to your team together with a conversation summary and customer details.
Enquiries, bookings and appointments
It collects details, checks availability, creates a booking or a CRM lead and sends the customer a confirmation, with no manual copying.
Many languages and scripts
It understands Montenegrin in Latin and Cyrillic, including text typed without diacritics, and replies to visitors in English, Russian, Turkish, German and other languages.
Conversation analytics
See the most common questions, unanswered topics and conversations handed to your team, so you know what customers want and where the knowledge needs filling in.
How it works
From a customer’s message to an accurate answer
Every message follows the same controlled flow, and you set the rules and limits.
The customer writes
The message arrives from your website, WhatsApp, Viber, Messenger or email into one system.
The model reads the intent
Our fine-tuned model understands the question, language and tone, and decides whether it’s a question, an order or a problem.
Knowledge base search
The system finds the relevant parts of your price lists, manuals and terms, or checks data in your CRM and booking system.
Answer or action
The chatbot answers from the sources it found, creates an enquiry or booking and sends the customer a confirmation.
Handoff when needed
Uncertain and sensitive conversations go to an operator with a summary, and everything is logged for analytics.
Use cases by industry
How an AI chatbot works in your industry
The same foundation, different jobs. We tune each chatbot to the questions and processes of your sector.
Tourism and hotels
A front desk that never sleeps
Guests get answers about rooms, rates and amenities in their own language, and booking requests land straight in your system.
- Availability and rates by date
- Transfer, parking and breakfast information
- Replies in English, Russian and Turkish
Real estate
Qualified leads for your agents
The chatbot understands what a buyer is looking for, suggests matching properties and books viewings, and the agent receives a lead with every detail.
- Listing search from a plain-language description
- Viewing scheduling
- Budget and contact details capture
Ecommerce
Sales assistant and support in one
It helps shoppers choose, answers delivery and returns questions and checks order status directly from your store.
- Product recommendations by need
- Order status and shipment tracking
- Delivery, exchange and return terms
Transport and ticketing
Routes, tickets and changes without the wait
Passengers check departures, fares and luggage rules, while change and refund requests reach operators already processed.
- Route and departure search
- Luggage and cancellation rules
- Email support with draft replies
Clinics and private practices
Appointments and preparation
Patients book appointments and get preparation instructions and price lists, while sensitive questions are passed to staff.
- Online booking and reminders
- Pre-visit preparation instructions
- Prices and practice opening hours
Banking, insurance and public services
Clear answers about procedures
Citizens and clients learn which documents they need and how a process works, on infrastructure that meets strict security requirements.
- Guidance through procedures and forms
- Product and terms information
- On-premise deployment for sensitive data
From our work
Chatbots and assistants we have built
We use the same approach in our own products and in client projects.
BusTicket AIClient project
AI customer support for bus ticket sales
AI support that answers emails and website chat in four languages, searches routes and passes uncertain cases to operators through a dedicated dashboard.
- FastAPI
- Spring Boot Modulith
- Next.js
- PostgreSQL
DreamEstate
AI platform for the real estate market
An AI assistant that understands what a property buyer wants in plain language and turns the conversation into a listing search.
- Spring Boot
- FastAPI
- Next.js
- PostgreSQL / pgvector
Kruna.ai
AI CRM for law firms
A specialist assistant that answers only from verified sources and shows where every piece of information comes from.
- Next.js
- Spring Boot
- RAG
- PostgreSQL
Packages and pricing
Choose the scope you need today
Start with a website chatbot and add channels and integrations as your business grows.
Website chatbot
from €800one-off + monthly maintenance
For businesses that want fast, accurate answers to their customers’ most common questions.
- Website chat widget
- Knowledge base from your documents
- Up to 2 languages
- Enquiries delivered by email
- Basic conversation analytics
Most popular
AI assistant
from €2,000one-off + monthly maintenance
For teams talking to customers on several channels who want enquiries automated.
- Website, WhatsApp, Viber, Messenger and email
- Model fine-tuned on your content
- Handoff to operators
- CRM or booking system integration
- Unlimited languages
- Advanced analytics and reports
Enterprise
from €5,000based on scope and integrations
For larger systems with internal databases, strict access rules and specific security requirements.
- Integration with multiple internal systems
- AI agent that completes tasks
- On-premise or dedicated infrastructure
- SSO, roles and audit log
- SLA and priority support
Prices are indicative and exclude VAT. The monthly cost of model hosting and maintenance depends on conversation volume and is set out in your written quote. All pricing
Comparison
Why not an off-the-shelf chatbot
Ready-made tools are quick to start with, but hit their limits as soon as the chatbot has to know your business.
| Criterion | ProCode AI chatbot | Off-the-shelf SaaS chatbot | Rule-based chatbot |
|---|---|---|---|
| Understands free-form questions | Yes | Partly | No |
| Answers only from your sources | Yes | Partly | Yes |
| Data stays on your or our EU infrastructure | Yes | No | Partly |
| Regional languages in Latin and Cyrillic | Yes | Partly | Partly |
| Integration with your CRM and bookings | Yes | Partly | Partly |
| Handoff to an operator with a summary | Yes | Partly | No |
| Knowledge base upkeep | We handle it | Your team | Writing rules by hand |
| Ownership of the solution | Yours | Subscription | Depends on the tool |
Process
A working chatbot in a few weeks
Clear phases, a fixed price per phase and testing on real questions before the chatbot ever talks to a customer.
Step 1 · 1 week
Workshop and enquiry analysis
We review your channels and most common questions and agree what the chatbot handles on its own and what it passes to your team.
Step 2 · 1–2 weeks
Knowledge base
We prepare and structure your price lists, manuals, terms and answers, and agree on the tone of voice.
Step 3 · 1–2 weeks
Model tuning and integrations
Fine-tuning on your content, connecting channels, your CRM or booking system, and setting handoff rules.
Step 4 · 1 week
Testing
We check accuracy against your customers’ real questions, fix weak spots and train your team.
Step 5 · ongoing
Launch and improvement
We switch the chatbot on across channels, monitor conversations and keep the knowledge up to date.
Technology
A fully in-house solution
We develop and run the models, the knowledge base and the infrastructure ourselves, so your conversations never pass through outside AI services.
Models and knowledge
- Fine-tuned language models
- RAG knowledge base
- Vector and hybrid search
- Intent recognition
- Language detection
Channels
- Website chat widget
- WhatsApp Business
- Viber
- Instagram and Messenger
Integrations
- REST API and webhooks
- CRM systems
- Booking systems
- Calendars
- Online stores
Infrastructure and security
- Private servers in the EU
- On-premise deployment
- SSO and roles
- Data encryption
- Audit log
Guide
What to know before adding an AI chatbot to your business
An AI chatbot can take over a large share of everyday customer communication, but only when it is set up well. This guide explains how an AI chatbot differs from a classic bot, how to prepare the knowledge base, what drives the cost and how to measure whether the chatbot is really helping your business.
What an AI chatbot is and how it differs from a classic bot
A classic chatbot follows predefined rules: the customer clicks buttons or types an exact keyword and the bot returns a prepared reply. As soon as a question falls outside the script, the conversation stalls. These bots are cheap to set up, but customers abandon them quickly.
An AI chatbot understands questions asked in the customer’s own words, with typos, without diacritics or in another language. Instead of a fixed script it uses a language model and your knowledge base to compose the answer. Connected to your systems, it can also get things done: check a slot, create an enquiry or find an order.
With us the model is always our own: fine-tuned on examples from your industry and running on private infrastructure. That means we can shape the chatbot’s behaviour precisely, and your customers’ conversations never end up with outside AI companies.
AI agents go a step further. While a chatbot mainly answers questions, an agent independently uses the tools you allow: it searches routes or listings, checks stock, fills in a form or drafts a quote. Every tool has clearly defined permissions, so the agent can never do anything that wasn’t agreed in advance, and every action is logged.
The knowledge base: the foundation of accurate answers
The most common reason for a bad chatbot is not the model but the knowledge it was given. A chatbot should answer only from verified sources, so we collect and organise the content with you before connecting it.
You don’t need huge amounts of data to start. The documents you already have are enough, and we structure them so the system can find and cite them easily.
The knowledge base isn’t a one-off job. Prices change, new services and seasonal offers appear, so your team can add or edit content through a simple panel without a developer. The system uses the new version of a document immediately and archives the old one, so the chatbot never quotes last season’s price.
- price lists, catalogues and service descriptions
- terms of business, delivery, cancellation and returns
- manuals, procedures and FAQs
- real customer questions and your team’s best answers
- live data from your systems, such as free slots or order status
Channels and integrations: where a chatbot adds the most
Customers pick channels out of habit, not to suit your organisation. Tourists often use WhatsApp, local customers Viber and Instagram, and business clients email. That is why we connect one assistant to every channel and keep conversations in one place, where your team can see and take them over.
The real value appears when the chatbot is more than an information source. Connected to your CRM, booking system, online store or calendar, it collects customer details, checks availability and creates an enquiry or booking. Your team receives complete, qualified requests instead of half-finished messages.
Security comes first when we connect systems. The chatbot only accesses the data it needs for a given task, never reveals one customer’s details to another, and reaches your systems through protected API keys with limited permissions. We test every integration in a staging environment before switching it on for real customers.
When a chatbot should hand over to a person
A good chatbot knows its limits. We agree in advance when a conversation goes to your team: complaints, special discounts, an unhappy customer, legally or medically sensitive questions, or simply when the model isn’t confident enough in its answer.
The operator doesn’t start from scratch. They get a summary of the conversation, the details the customer has already shared and a suggested reply, so they can respond in seconds. For email support, the chatbot can prepare draft replies that staff just review and send.
Tone matters just as much. The chatbot speaks in your brand’s voice: formal for a bank or law firm, warmer and more relaxed for a hotel or restaurant. We define tone, answer length and phrases to avoid during the workshop and build them into the model, so customers feel they are talking to your company, not to a generic program.
How much an AI chatbot costs and what drives the price
The price depends on the number of channels, the size of the knowledge base, the number of integrations and security requirements. A website chatbot that answers common questions and captures enquiries starts at €800. A multichannel AI assistant with operator handoff and CRM or booking integration starts at €2,000. Enterprise solutions with several internal systems and on-premise deployment start at €5,000.
On top of development there is a monthly cost for model hosting and maintenance, which depends on conversation volume. After the workshop you receive a written quote with a fixed price per phase and no hidden per-seat fees.
How to measure whether your chatbot is working
A chatbot isn’t a project that ends at launch. In the first weeks we review conversations and fill gaps wherever questions went unanswered. Analytics show which topics matter most to customers, which is useful for sales and for improving your service.
We recommend a monthly results review with your team. Based on the analytics, we agree which topics to add to the knowledge base, which process to automate next and whether the chatbot should expand to new channels. That way the solution grows with your business instead of going stale a few months after launch.
- share of conversations resolved without an operator
- time to first reply across all channels
- number of enquiries, bookings and appointments captured
- topics and questions the chatbot couldn’t answer
- customer ratings of conversations
FAQ
AI chatbot FAQs
Answers to the questions businesses ask us most often before launching a chatbot.
How much does an AI chatbot for a business cost?
A website chatbot that answers from your knowledge base and captures enquiries starts at €800. A multichannel AI assistant (WhatsApp, Viber, Messenger, email) with operator handoff and CRM integration starts at €2,000, and enterprise solutions connected to several internal systems start at €5,000. There is also a monthly cost for model hosting and maintenance that depends on conversation volume. After a workshop you get a written fixed-price quote.
Which AI models do you use?
We use our own language models, fine-tuned for each client on examples from their industry and on their own content. The models run on our private infrastructure in the EU or on your servers. Your data and your customers’ conversations are never sent to outside AI companies and are never used to train models for other clients.
Can the chatbot understand Montenegrin and other regional languages?
Yes. Our models are adapted to the region’s languages and understand Montenegrin, Serbian, Bosnian and Croatian in Latin and Cyrillic, including informal writing without diacritics. The chatbot replies in the language the customer uses, so it can talk to international visitors in English, Russian, Turkish or German at the same time. Industry terminology is added through the knowledge base.
How do you stop the chatbot from making up answers?
The chatbot answers only from your knowledge base and data from your systems. When information is missing, it is set up to say so and offer a contact with your team rather than guess. Before launch we test it against your customers’ real questions, and after launch we monitor conversations and fill in knowledge wherever it is needed.
Can the chatbot work on WhatsApp, Viber and Instagram?
Yes. We connect one assistant to your website, WhatsApp Business account, Viber, Instagram, Facebook Messenger and email. Customers get the same information on every channel, and your team sees every conversation in one place. We take care of the technical setup, business account verification and all connections.
What happens when the chatbot doesn’t know the answer?
It hands the conversation to your team together with a summary and the details the customer has already provided. We also agree in advance which situations always go to a person, such as complaints, special offers or an unhappy customer. Outside working hours the chatbot takes contact details and tells the customer when to expect a reply.
Is the chatbot GDPR compliant, and where is data stored?
Yes. Data is stored on our servers in the EU or on your own infrastructure, with encryption and access control. Customers are told they are talking to an AI assistant, we collect only the data needed for the enquiry and set a retention period for conversations. For banks, clinics and the public sector we offer on-premise deployment where nothing leaves your network.
How long does it take to build an AI chatbot?
A website chatbot with a prepared knowledge base can be ready in 2 to 3 weeks. A multichannel assistant with integrations usually takes 4 to 6 weeks, and enterprise projects take longer depending on the number of systems. Most of the time goes into organising content and testing against real questions, because that is what determines answer quality.
Can the chatbot book appointments and take orders?
Yes, once it is connected to your booking system, calendar, CRM or online store. The chatbot then checks available slots, collects the required details, creates a booking or enquiry and sends the customer a confirmation. If your system has no API, we propose the safest way to exchange data before any work begins.
Who maintains the chatbot after launch?
We do. We monitor answer quality, update the knowledge base when prices or terms change, improve the model and look after infrastructure and security. Your team can also edit content through a simple panel, and every month you receive a report on conversations and unanswered topics.
AI for business
Specialised AI solutions
Show us the questions your customers ask most
In a short free consultation we estimate how many conversations a chatbot can take over, which channels make sense and what the solution would cost.