What an AI chatbot can do today
AI chatbots today are built on large language models (LLMs). That means no more rigid decision trees, no predefined answer paths. Instead, a modern chatbot understands context, holds natural conversations, and responds flexibly to what visitors actually ask.
For Swiss SMEs, this represents a real lever. A well-configured AI chatbot does more than answer common questions — it qualifies leads, detects purchase intent, and guides visitors purposefully. The technology is mature. According to current figures, over 60% of B2B companies already use chatbot software. And Gartner forecasts that by 2027, one in four companies will use a chatbot as their primary customer channel.
The question is no longer whether an AI chatbot makes sense. The question is which approach fits your business.
Three use cases beyond FAQ
Answering FAQs is the most obvious use case. But an AI chatbot can do significantly more — when the conversation logic is right.
Lead qualification
Appointment booking
Onboarding & consulting
SaaS chatbots: Fast start, clear limits
Platforms like JotForm AI Agents, Tidio, or Chatfuel make getting started simple. Without any coding knowledge, you can set up a chatbot in a few hours that answers common questions and routes visitors onward.
For pure FAQ bots, this is often sufficient. JotForm also offers features like Human Takeover, over 200 integrations, and a voice mode. Entry costs are manageable: from around CHF 40 per month.
The limits show up as soon as the chatbot needs to deliver more than standard answers. Specifically at three points:
Tone and brand voice. SaaS platforms offer basic personalisation — tone of voice, a few adjectives, a short description. For a consistent brand voice across different conversation situations, that rarely suffices.
Instruction-following. With more complex conversation flows — for example when the chatbot should first qualify and then book — SaaS bots frequently deviate from instructions. This leads to inconsistent experiences.
Integration depth. Many integrations exist on paper. In practice, this often means an embed link to an external tool rather than a native integration within the conversation flow.
SaaS chatbots are a good starting point. For pure FAQ answering and simple routing, they are often adequate. The decision for custom becomes relevant when conversation quality and integration depth determine whether visitors become leads.
Why instruction-following makes the difference
Instruction-following — how reliably a chatbot follows directives — sounds like a technical detail. In practice, it determines whether a conversation becomes a lead or breaks off.
A concrete example: for noevu.ch, JotForm AI Agents was tested first. The platform offers impressive features. But even after several rounds of adjustments, reliable conversation steering proved impossible. The bot deviated from instructions, mixed tones, and could not sustain a structured lead qualification.
That is why a custom chatbot was developed — noevu.ai. With a detailed instruction system that sets priorities: appointment booking takes precedence over informational conversation. One question per response, maximum 60 words. Confirmation before follow-up question. These rules hold reliably — because they are anchored directly in the system, not configured through a generic interface.
The result: a chatbot that speaks in the brand voice, runs structured conversations, and brings visitors directly to an appointment. No detour through an external tool.
The biggest insight: a chatbot that first confirms and then asks converts better. "Yes, certainly — one quick question" builds trust. SaaS platforms rarely allow you to control this with that level of precision.
SaaS vs. Custom: The honest decision guide
Both approaches have their place. What matters is what the chatbot should do for your business. This overview shows the key differences.
| SaaS chatbot | Custom chatbot | |
|---|---|---|
| Setup time | Hours | 3–5 working days |
| Cost (monthly) | From ~CHF 50 | LLM API costs, scaling |
| Instruction-following | Limited control | Precisely controllable |
| Brand voice | Limited customisation | Brand-aligned |
| Calendar integration | External embed | Native in chat |
| Multilingual | Automatic, limited | Configured per language |
| Data ownership | With provider (often US) | Freely selectable |
| Conversation logic | Pre-built flows | Freely definable |
Prices as a guide in CHF, as of April 2026. Custom: one-time development cost, ongoing = LLM API costs only. Binding figures come with an individual quote.
What a custom chatbot can do in practice
A custom chatbot is not a third-party tool embedded in the website — it is an integrated part of it. The noevu.ai example shows what that means in practice.
Features in detail
Development took around four days — from first requirements to production. No third-party platform, no lock-in. The chatbot is fully part of the website stack and can be adjusted at any time.
This approach adapts to your business — built around your existing systems, in your brand voice, under your data ownership. Find out more at Contact.
n8n connects tools, automates processes, and complements AI chatbots perfectly. The practical check for Swiss SMEs.
Which integrations are worth it
A chatbot without integration is an isolated tool. The real leverage comes where the chatbot works together with existing systems.
Calendar (Cal.com, Calendly): Show available slots in the conversation and book directly — for service businesses, the biggest single lever. No media break, no external form.
CRM (HubSpot, Bexio, Salesforce): Transfer qualified leads automatically to the CRM. Name, email, topic, and conversation context — without manual effort.
Products & inventory: Retrieve current stock levels, prices, or product details directly in the conversation — instead of sending visitors to an overview page.
Website navigation: Configure the chatbot to point to specific pages, offers, or documents — contextually, not generically.
Messaging & notifications: Notify teams of urgent enquiries via Slack or email. Send booking confirmations or reminders directly to customers.
Knowledge base: FAQs, product information, and company details as structured data. The better the base, the more precise the answers. Can also be kept up to date automatically with automation tools like n8n.
The principle is always the same: if a system has an open interface (API), it can be connected.
Integration without clear conversation logic creates chaos. Define the conversation goals first, then connect the integrations. Otherwise the chatbot books appointments before it has understood the topic — and the team receives context-free bookings.
Checklist: Does an AI chatbot fit your SME?
Not every SME needs an AI chatbot. This checklist helps with the assessment. The more points apply, the greater the potential benefit.
Assessment: With 7 or more points, an AI chatbot is a promising next step. With 4–6 points, a closer look is worthwhile — a simple SaaS solution may be enough. With fewer than 4 points, there are probably more effective levers available. Which AI use cases are relevant for your SME is covered in the overview article.
Summary: Invest now, build a competitive edge
AI chatbots are no longer an experiment. They are a tool that delivers measurable results today — from lead qualification to appointment booking.
The real question is not whether an AI chatbot makes sense. It is which approach fits. SaaS platforms are a good starting point for simple use cases. For businesses that need more — a consistent brand voice, deep integrations, real conversation guidance — the custom approach pays off.
Those who start now build a cumulative advantage. Every conversation delivers data for better instructions. Every iteration improves the conversion rate. That lead is hard to close later.
Switzerland has catch-up potential in AI adoption, according to the Switzerland AI Handbook by digitalswitzerland. That is exactly where the opportunity lies: SMEs that act today position themselves ahead of the competition — not behind it.
Whether FAQ bot or sales assistant — a short conversation clarifies the next steps. Concrete, no obligation, and specific to your situation.
Frequently Asked Questions
What does an AI chatbot cost for a Swiss SME?
Costs depend on scope. SaaS platforms start at around CHF 50 per month, but offer limited control over tone and integrations. Custom chatbots are higher, depending on integration depth, languages, and conversation logic. From around CHF 2,000 setup plus ongoing operating costs is a realistic benchmark. A binding quote comes after the initial conversation.
How quickly is an AI chatbot ready to go?
SaaS platforms are up in hours. A custom chatbot with instruction development, FAQ preparation, integration, and testing takes around 3 to 5 working days. Depending on integration depth — such as calendar connection or CRM synchronisation — it may take slightly longer.
Can an AI chatbot book appointments directly?
Yes. With a calendar integration (e.g. Cal.com or Calendly), the chatbot shows available time slots directly in the conversation. Visitors book without leaving the page. On noevu.ai, the chatbot first collects name, email, and topic — then the slot selection opens.
What integrations are possible with an AI chatbot?
Essentially any system with an open interface (API). Typical examples: Cal.com or Calendly for appointments, Bexio or HubSpot for CRM, Slack or Teams for escalation to the internal team. Integration depth depends on existing systems — not on the chatbot platform.
Does an AI chatbot work in multiple languages?
Yes. Modern AI models handle dozens of languages. What matters is that the instructions and knowledge base are maintained in each language. A custom chatbot detects the language version of the website and responds accordingly — with language-specific FAQs and conversation rules.
What happens to conversation data?
That depends on the solution. With SaaS platforms, data often sits on US servers. With custom solutions, you control hosting and data flow. Since the revDSG (September 2023), Switzerland has strict transparency requirements: visitors must know they are talking to an AI, and data processing must be documented.
When is a SaaS chatbot enough — and when does Custom become necessary?
SaaS is enough when the chatbot only answers common questions and needs no deep integrations. Custom pays off as soon as the chatbot should qualify leads, book appointments, or speak in the brand's tone. The line is where instruction-following and integration depth determine the outcome.
