FORTALAN Chatbots, custom AI automations and voice agents. Zagreb, Croatia. This is the plain-text version of the site: the home page and the three direction pages, generated from the pages themselves. Every demo on this site is recorded on fictional data (sample data). None of it is a live system. PAGES Home / Chatbots /chatbots/ Custom AI automations /automations/ Voice agents /voice/ Privacy notice /privacy/ Legal notice /legal/ How our demos are made /demo-notice/ ================================================================================ HOME PAGE / ================================================================================ WHAT WE BUILD AI assistants for websites, teams and phone lines. Built on your own data, tested before launch and reviewed after it. - Chatbots: A website assistant that knows your catalogue and guides customers to the right product. - Custom AI automations: AI agents that take routine work off your team, inside the tools it already uses. - Voice agents: A voice agent that answers calls and books appointments, in English and Croatian. CHATBOTS: THE USED-CAR DEALER EXAMPLE, WITH SAMPLE DATA Used-car dealer A used-car dealer’s website. The customer describes the car they want; the assistant finds three that fit, opens the chosen one and sends the test-drive request to sales. 1. The shopper describes the car in plain words and answers one question on icon chips: diesel or hybrid. 2. In the side panel, the assistant sets the site’s own filters, 37 cars become 4, and it picks three with a reason each. 3. The shopper picks one, the assistant opens that car’s page, and the test-drive request goes to the sales team with Send. Video, no sound: /media/chat-dealer-v2-home.mp4. The steps above say what it shows. Sample data: a fictional dealer and a scripted demo, sped up in places. More about chatbots (/chatbots/) CUSTOM AI AUTOMATIONS: THE IT HELPDESK EXAMPLE, WITH SAMPLE DATA IT helpdesk The IT helpdesk of a wholesale distributor. When an email arrives, the AI agent resolves it or tells the IT team exactly what to do. 1. A user emails the helpdesk, and the email becomes a ticket. 2. The agent checks past tickets and the knowledge base and finds the same problem, already solved. 3. Known fix: it replies and resolves the ticket; needs IT: it alerts the helpdesk team with what to do. Video, no sound: /media/auto-resolver-home.mp4. The steps above say what it shows. Sample data: a fictional company and a scripted demo, sped up in places. Its emails and tickets were written for this demo. More about custom AI automations (/automations/) VOICE AGENTS: THE PHYSIOTHERAPY CLINIC EXAMPLE, WITH SAMPLE DATA Physiotherapy clinic A physiotherapy clinic. A patient calls, the voice agent books the appointment, and it appears in the clinic’s calendar. 1. A patient calls about a sore knee and asks for Thursday, around five. 2. Thursday 17:00 is taken, so the agent books 18:00 and puts her on the waitlist for 17:00. 3. The clinic’s dashboard shows the call with its summary, and the booking is already in the calendar. Video, no sound: /media/voice-reception-home.mp4. The steps above say what it shows. Sample data: a fictional clinic and a scripted call, shortened. The call is in English; the agent also takes calls in Croatian. More about voice agents (/voice/) HOW WE WORK Three stages, one process. You can stop after any stage, and you stay in control throughout. From a free demo on your own site to a live assistant under regular review. Every project follows the same stages, with clear responsibilities on both sides. 1. Demo Free. A short private video of the assistant on your own site, usually within a few days. Your part: Send your website’s address. Our part: Build it from your public pages and record it answering questions about your products. 2. Pilot It goes on your site, your phone line or into your team’s tools for an agreed scope. You decide what it may answer, what it may do on its own and where it hands over. Your part: Your data, and someone to take the handovers. Our part: Set-up, its own tests, and a written list of failures with the guard for each. 3. Live We regularly read conversations and fix weak answers. Your part: Tell us when something changes, such as a new returns rule or a new service. Our part: Regular reviews and fixes. You own the data and the off switch. Every build comes with its own tests. Questions it must decline, live-site checks, a written list of failures and the guard for each. Prices, delivery dates and policies come from your data, and when the data is missing, it says so. Case | Check | What happened | Result Used-car dealer | Asks for a price the stock list does not show | Declined, offered the sales team | Pass Used-car dealer | Sets body, fuel and gearbox filters on the page | 3 of 3 set | Pass Used-car dealer | Asked for a delivery date the data does not hold | Promised a date; guard added to the set | Fail Support line | Caller cannot be verified | No ticket, transferred to a person | Pass Ticket resolver | Ticket with no known fix | Handed to a person with what it found, no reply sent | Pass Sample run, fictional data. Every failure stays in the set as a permanent case, with its guard. CASE STUDIES A selection of recent builds. Client names are withheld. 1. Shopping assistant for a used-car dealer (Chatbots, Retail). - Covers the dealer’s entire live stock, about 300 cars in our snapshot - Tested turn by turn on the live site 2. Website assistant for a telecom operator (Chatbots, Telecom). - Built from about 660 pages and close to 500 PDFs - Answers in about two seconds in our recorded run 3. Voice incident intake for business ICT services (Voice agents, ICT services). - Takes the call in Croatian - The ticket appears on the engineers’ board within about 15 seconds in our demo 4. Voice receptionist for appointment-based businesses (Voice agents, Services). - Books within working hours and capacity - Offers a waitlist when a slot is taken 5. In-app help for business software (Chatbots, Software). - Answers from nearly 2,000 pages of manuals - Declined every control question the manuals could not answer: 12 of 12, a small control set Figures are rounded and come from our own builds and recorded runs. We show no logos and no client quotes without written permission. ABOUT An independent studio in Zagreb, working in English and Croatian with businesses in any country. I started Fortalan in 2026 to build assistants that know where their knowledge ends. Each one starts from your own pages, tickets or manuals, comes with its own tests, and stays under your control. You deal directly with the people who build it, from the first demo to the regular reviews. Founder, Fortalan Location: Zagreb, Croatia Languages: English and Croatian Time zone: Central European Time Reply: Within 2 working days Legal: A Croatian sole proprietorship (obrt). Legal notice (/legal/) Principles behind every build - It tells people they are talking to an AI assistant. - When it is unsure, it says so and offers a way to reach your team. - Prices, delivery dates and policies come from your data, and when the data is missing, it says so. - We regularly read conversations and fix weak answers. - You own the data and the off switch. - Every build comes with its own tests. A DEMO ON YOUR SITE We record a short, private video of the assistant working on your own website. Free of charge, usually ready within a few days. Or write directly hello@fortalan.com (mailto:hello@fortalan.com) We reply within 2 working days, in English or Croatian. The demo uses public pages only and needs no access to your systems. The video is sent to you and nobody else. The contact form asks for: Your name; Email; Website address; What do your customers ask most? Optional. The form needs JavaScript to open your mail app: write to hello@fortalan.com instead. ================================================================================ CHATBOTS /chatbots/ ================================================================================ AN ASSISTANT THAT KNOWS ITS WAY AROUND YOUR CATALOGUE. We make the website live and easier to navigate. HOW IT WORKS 1. The shopper writes in plain words. 2. The assistant reads your catalogue, pages and policies. 3. It sets your site’s own filters and shows products with a reason for each. 4. It asks one question when the answer changes what fits. 5. It opens the product page or the filtered listing. 6. When a person is needed, it hands over to your team with the whole conversation. It says it is an AI, and it says when data is missing. What the assistant does between the shopper and your site, step by step. USED-CAR DEALER Wendlecote Motors, a used-car dealer: a shopper describes the car in plain words, and the assistant finds three that fit, opens the right pages and passes the test-drive request to sales. Fictional customer. The problem: Shoppers know what they need, but not which of the many filters to set, so they scroll and leave. What it does, in three steps 1. The shopper says what they need. The shopper docks the assistant as a side panel, and the page makes room. Then the shopper describes the car in plain words and answers one question on icon chips: diesel or hybrid. 2. It sets the filters, picks three. The assistant sets the site’s own filters, 37 cars become 4, and it picks three with a reason each. 3. Opens the right pages, sends to sales. The shopper picks one, the assistant opens that car’s page, and the test-drive request goes to the sales team with Send. Asked about the battery warranty, it opens the Warranty page at the highlighted sentence. Video, no sound: /media/chat-dealer-v2.mp4. The steps above say what it shows. Sample data: Wendlecote Motors is a fictional dealer. Its stock is fictional, with open-licence photos credited on the legal notice. The demo is scripted and sped up in places. Tested with: Questions it must decline, such as a price or a delivery date the stock list does not hold, and test turns in the browser that check the page’s own filters after every answer. Built from: A catalogue index, a control for the page’s own filters, and a handoff to the sales team. HOW OUR CHATBOTS WORK The same rules on every website we build for. What it does: - Finds products from a plain description - Sets the filters and parameters on the page - Answers from your catalogue, pages and policies - Asks one question when the answer changes what fits Prices, delivery dates and policies come from your data, and when the data is missing, it says so. Handoff: It hands over with the whole conversation, by your contact form, helpdesk or email, or by live chat. It tells visitors they are talking to an AI assistant, and when it is unsure, it says so and offers a way to reach your team. Tested with: Every build comes with its own tests: questions it must decline, live-site checks after every answer, and a written list of failures with the guard for each. We regularly read conversations and fix weak answers. Built from: Your catalogue, pages and policies as its sources, a control for your page’s own filters and parameters, and a handoff to your team. You own the data and the off switch. A DEMO ON YOUR OWN SITE A short private video of the assistant on your own pages. It is free, and usually ready within a few days. Ask for a free demo (/#contact) We reply within 2 working days, in English or Croatian. ================================================================================ CUSTOM AI AUTOMATIONS /automations/ ================================================================================ ASSISTANTS BUILT AROUND YOUR TEAM’S OWN WORK. We automate the work your team repeats. HOW IT WORKS 1. Something comes in to your own tools: an email, a ticket, a new build or a changed page. 2. The agent reads your sources (tickets, the knowledge base, the app, the manuals) and works inside your own tools. 3. Where you allow it, it finishes the task itself: it replies and resolves the ticket, runs the test or updates the page. Everywhere else, it hands the task to a person, with what it found. Every step is logged with its sources. It starts read-only; writing is a switch you control. What the agent does with each task, step by step. The three cases below show it on a helpdesk, a test suite and a manual. TICKET RESOLVER Dunmarrow, a wholesale distributor’s IT helpdesk: an email comes in, and the AI agent solves it or tells IT what to do. Fictional customer. The problem: The same questions reach the desk every week, and each one waits in the queue until someone has time. What it does, in three steps 1. An email arrives and becomes a ticket. A user emails the helpdesk, and the email becomes a ticket. 2. Checks past tickets and the knowledge base. The agent checks past tickets and the knowledge base and finds the same problem, already solved. 3. Known fix: resolved. Needs IT: alerted. Known fix: it replies and resolves the ticket; needs IT: it alerts the helpdesk team with what to do. Video, no sound: /media/auto-resolver.mp4. The steps above say what it shows. Sample data: Dunmarrow is a fictional company. Its emails, tickets and knowledge base were written for this demo. The demo is scripted and sped up in places. Tested with: Tickets it never saw while it was built, with every answer graded against the sources it cited, and tickets that look familiar but need a person, which it must not answer on its own. Built from: An index of the knowledge base and past tickets, a list of verified fixes that may go out on their own, an alert that hands the rest to the helpdesk team with what it found, and a helpdesk connection with the rights you set. END-TO-END TESTING ASSISTANT Benchwick, a software company: before a release, the assistant tests its booking app like a customer and reports what broke. Fictional customer. The problem: Before every release, someone clicks through the same flows by hand, and a broken step slips through when time is short. What it does, in three steps 1. A new build, four test scenarios. A new build is on staging, and the assistant queues four test scenarios in two browsers. 2. Books like a customer, checks each step. It books a workshop like a customer, checks every step, and finds the discount applied twice. 3. A report with screenshot and suggested fix. It writes the report: 3 passed, 1 failed, with a screenshot, expected and actual, and a suggested fix. Video, no sound: /media/auto-e2e.mp4. The steps above say what it shows. Sample data: Benchwick is a fictional software company, and the app under test is a demo app. The test run is scripted and sped up in places. Tested with: A test copy of the app with faults planted in it on purpose, which it must find, and flows that work, which it must pass without false alarms. Built from: A browser it drives step by step, test scenarios written in plain language, checks against what the screen shows, and a report with screenshots. DOCUMENTATION ASSISTANT Spokeline, a business-software vendor: a gap in its manual becomes a new section, drafted from the app and approved by a person. Fictional customer. The problem: The app changes every month, and the manual still describes screens that no longer exist. What it does, in three steps 1. The manual misses: cancelling a booking. The gap list shows what the manual never covered, and a person picks one: cancelling a booking. 2. Walks the app, drafts with screenshots. It walks through the feature in the app, like a person, and drafts the section with fresh screenshots. 3. A person approves, the manual updates. A person approves the draft, and the manual gets the new section: 4.3.1 Cancelling a booking. Video, no sound: /media/auto-docs.mp4. The steps above say what it shows. Sample data: Spokeline is a fictional business app, and Larkmoor Cycles a fictional bike shop that uses it. The demo is scripted and sped up in places. Tested with: Machine checks on every draft before a person reads it: the steps match the screens, every button name is on screen, and every picture has a caption that matches it. Built from: A browser session with a step ledger, a section writer, checks against the screen, and a diff with undo. HOW OUR AUTOMATIONS WORK Built around the tools your team already uses, with a person in charge of what matters. What it does: - Works inside your own tools: the helpdesk, the web app, the manual - Acts on its own only where you allow it; everywhere else, it hands the task to a person with what it found - Keeps a log of every step, with its sources and screenshots - Starts read only; writing is a switch you control Handoff: A person stays in charge. Whatever it may not finish on its own, it hands to the right person on your team with what it found, and a new manual section waits in a review step until a person approves it. Tested with: Every build comes with its own tests: cases from your own history that it never saw while it was built, faults planted on purpose that it must catch, and a written list of failures with the guard for each. Built from: Your knowledge base, tickets, app and manual as its sources, a browser it can drive, connections with the rights you grant, and a review step with undo. You own the data and the off switch. TELL US WHICH WORK REPEATS Describe the tickets, test runs or manuals that take your team’s time, and what a person should keep deciding. Ask for a demo (/#contact) We reply within 2 working days, in English or Croatian. ================================================================================ VOICE AGENTS /voice/ ================================================================================ A PHONE LINE THAT PICKS UP, IN ENGLISH AND CROATIAN. Our voice agents answer every call, take the details, read them back and leave a clean record: a booking in your calendar or a ticket with the transcript. HOW IT WORKS 1. A call comes in, in English or Croatian, at any hour. 2. The agent verifies the caller when needed, takes the details and reads them back. 3. It books into the calendar (or offers the waitlist), or opens a ticket. 4. The transcript and a summary go to your team. It transfers to a person when the call needs one. What the voice agent does with each call, step by step. RECEPTIONIST Tansywell Physio, a physiotherapy clinic: a patient calls, the voice agent books the visit, and it lands in the calendar. Fictional customer. The problem: The phone rings while every therapist is with a patient, and a missed call is a missed booking. What it does, in three steps 1. A patient calls to book a visit. A patient calls about a sore knee and asks for Thursday, around five. 2. 17:00 taken: books 18:00, waitlist for 17:00. Thursday 17:00 is taken, so the agent books 18:00 and puts her on the waitlist for 17:00. 3. Summary and booking in the dashboard. The clinic’s dashboard shows the call with its summary, and the booking is already in the calendar. Video, no sound: /media/voice-reception.mp4. The steps above say what it shows. Sample data: Tansywell Physio is a fictional clinic, and the call is scripted between fictional parties and shortened. The call is in English; the agent also takes calls in Croatian. Tested with: Calls that must not end in a booking: outside working hours, on a full day, or with a caller who changes their mind; and every booking checked against the calendar after the call. Built from: A voice agent, the clinic’s calendar with working hours and capacity, a waitlist, and a dashboard with each call’s transcript and summary. SUPPORT LINE Holtwick, a facilities maintenance company: a caretaker calls about a leak, and the agent puts a work order on the board. Fictional customer. The problem: Calls about a stuck lift or a leak come in at any hour, and details get lost between the call and the job. What it does, in three steps 1. A caretaker calls about a leak. A caretaker enters his site code and reports water coming through the car park ceiling. 2. Asks, reads back, opens a work order. The agent asks follow-up questions, reads the problem back and opens a work order once he confirms. 3. On the board: P1, crew, transcript. The work order lands on the board with priority P1, the plumbing crew, a deadline and the transcript. Video, no sound: /media/voice-support.mp4. The steps above say what it shows. Sample data: Holtwick is a fictional company, and the call is scripted between fictional parties and shortened. Tested with: Calls that must not open a work order: a caller who cannot be verified, a building the company does not look after, or too few details to act on. Built from: A voice agent, caller verification, a work order service with priorities, and the maintenance team’s board with the transcript. HOW OUR VOICE AGENTS WORK The same rules on every line we build for. What it does: - Answers in English or Croatian - Verifies the caller when the call needs it - Takes the details and reads them back - Books into your calendar, or opens a ticket with the transcript Handoff: It transfers the call to a person when the call needs one, and every call it takes leaves a transcript and a summary for your team. It tells callers they are talking to an AI assistant. Tested with: Every build comes with its own tests: scripted test calls with wrong details, full days, unclear answers and callers who switch language, and every booking and ticket checked after the call. Built from: A voice agent in English and Croatian, your calendar or ticket system, caller verification where the call needs it, and a dashboard with calls, transcripts and summaries. You own the data and the off switch. TELL US WHAT YOUR CALLERS NEED Bookings, support calls or both: describe your line and what it should handle. Ask for a demo (/#contact) We reply within 2 working days, in English or Croatian. ================================================================================ Fran Talan IT, obrt za računalno programiranje, vl. Fran Talan (OIB 91576422493, Sesvete, Zagreb) Every demo shown on this site is recorded on fictional data.