AI client support
Answers common questions online around the clock – about deliveries, terms and products – and hands over to a person with the full conversation when needed.
An AI chatbot that answers from your own documents
Not a chatbot that guesses, but one that reads your manuals, contracts and articles – and tells you when it doesn't know.
Customer service on the websiteExample
Connected to what you already have
The chatbot answers from your existing sources, so no one needs to rewrite content for it. When a source is updated, its answers are too, without anyone feeding it text manually. It also works with Notion and most systems that have an API.
Your website
Pages, product sheets and FAQs. The chatbot reads the same content as your visitors and answers wherever they ask.
SharePoint
Handbooks, procedures and policies in SharePoint or Google Drive. Existing permissions carry over, so each employee only gets answers from content they already have access to.
CRM
Client history and previous cases in your CRM or case management system provide answers based on each client's situation, not just your general terms.
Teams
The internal assistant is available in Teams or on your intranet, where employees already work, so no one needs to open a new tool to ask a question.
For your clients or your team
Internal AI assistant
Employees ask questions and get answers from handbooks, procedures and policies, with the source alongside. New starters get up to speed faster.
Product and sales enquiries
Helps visitors find the right product or service for their needs, and passes qualified enquiries on to sales.
Support for technicians
Gives service technicians and support teams quick answers from technical manuals and previous fault reports, even on their phones out in the field.
How does the chatbot know what's correct?
The technology is called RAG – retrieval augmented generation. Every answer goes through four steps.
- 01
A question comes in
Someone types a question in their own words, in Swedish or another language – in a chat on your website, in Teams or on your intranet. The chatbot doesn't need keywords or predefined buttons. It also understands spelling mistakes, follow-up questions and questions phrased in ways no one anticipated.
- 02
Searching your sources
Before the language model responds, the chatbot finds the passages in your documents and systems that relate to the question: product sheets, terms, procedures or previous cases. Sources are indexed in advance and updated when you make changes, so answers reflect what's current, not what the model happens to remember.
- 03
An answer with a source
The language model puts together an answer based on what was found and shows the source behind it. It is instructed to stick to your documents rather than guess. Users can click through to the source and check for themselves, and you can review which sources each answer was based on.
- 04
Handing over
If there is no reliable answer, the chatbot says so and offers to connect the user with a person. The conversation is passed on during the handover, so the client doesn't have to repeat themselves. Questions the chatbot couldn't answer are also collected, showing which documents are missing or need to be clearer.
Not every question needs a chatbot
A chatbot is useful when the same questions are asked in many different ways and the answers are in your documents. If you can show the answer directly on the page, that's often better. And if a task follows the same steps every time, a standard workflow is enough.
- Answer questions about deliveries and returns around the clock/ AIThe same questions, phrased differently. This is where a chatbot helps.
- Find the right procedure in a 200-page staff handbook/ AIThe answer is there, but no one can find it. The chatbot finds it for them.
- Send a confirmation when a support ticket is created/ AutomationThe same email every time. A standard workflow is enough.
- Display opening hours and phone numbers/ NothingMake them easy to find on your website. No one should have to ask.
You decide what it can answer
The chatbot stays within the boundaries you set: which topics it answers questions about, which sources it can use and when it should hand over. After launch, we check that it follows those rules and make adjustments when needed.
01Topics and sources
The chatbot only answers questions on the topics you choose, using the sources you specify. For questions outside that scope, it politely declines and points the user in the right direction.
- You choose the topics and sources
- Users are directed elsewhere for questions outside its scope
- The source is shown with every answer
02Client and staff access stay separate
A client-facing chatbot and an internal assistant have different sources and permissions, so internal documents never appear in an answer to a client.
- Separate sources for clients and staff
- Permissions from your existing systems
- Personal data excluded where it isn't needed
03Sensitive questions go to a person
Complaints, legal matters, health or finance – you decide which topics the chatbot should never answer on its own. It then hands over, along with the full conversation.
- Topics that always trigger a handover
- The conversation is passed on at handover
- No guessing when there is no answer
04Monitoring what it does
We track how many questions the chatbot resolves on its own, how often it hands over and which questions it can't answer. That last list is often the most valuable: it shows exactly what information is missing from your website or procedures.
- Proportion of questions resolved directly
- Questions handed over
- A list of missing information
What is an AI chatbot for business?
An AI chatbot for business is a chat tool built on a language model that answers from the company's own sources – manuals, contracts, articles and previous cases – rather than what the model happens to know. It understands questions in free text, answers in Swedish and shows where the answer comes from.
It differs from older chatbots with buttons and decision trees, which can only handle questions someone has anticipated. It also differs from an AI agent: a chatbot answers, an agent acts.
Frequently asked questions about AI chatbots
Can't find the answer? Book an AI review or call us on 070-953 53 99.
AI review
30 minutes, one of your workflows
We review a task that takes up time today and give you an honest view of what fits: AI, standard automation or no change at all.
- A brief written proposal for next steps afterwards
- Meet the people who will build it
- Free of charge
