AI chatbots

Business chatbots: good support goes beyond the answer

We build chatbots connected to your CRM, orders and customer support tools. They check order status, gather details for service visits and open support tickets. When a support agent takes over, they have the context to continue without asking the customer to start again.

What is a business chatbot?

A business chatbot is an assistant that customers or colleagues can talk to. We build AI chatbots that can access information in your existing CRM, business software, helpdesk and company documents. This lets them answer questions, carry out agreed tasks and hand the conversation to a support agent when needed.

The problems to solve

The customer has to start again

The chatbot keeps replying without resolving the issue. When a support agent gets involved, the customer has to explain everything again. Both end up repeating work already done.

It cannot get beyond FAQs

It can give opening hours and the warehouse address, but it cannot check whether an order has shipped, book a service visit or say when the technician will arrive. Your team still has to handle those requests manually.

Nobody keeps the information current

Prices, procedures and opening hours change, but nobody is responsible for updating the information the chatbot uses. It keeps giving answers that used to be correct.

What we build

Connections to your CRM, orders and helpdesk
We connect the chatbot to the systems you use, including Odoo, so it can read order status, availability and customer history directly from the source.
Clear rules for when to stop
We agree which requests the chatbot can handle, which topics need a person and how many attempts it should make before handing over.
A handover with the context your team needs
The handover includes who the customer is, what they need, what has already been checked and what information is still missing. The support agent can pick up from there without asking the customer to repeat their explanation.
A clear maintenance plan
We document who updates the information, how often and how answers are checked. That work is part of the project too.

What we do not do

We do not promise an answer to everything
Before building the chatbot, we agree what it should handle. Requests outside that scope are passed to a person.
We do not assume you need a chatbot
If you receive few requests and each needs a different judgement call, a chatbot may offer little value. We work through that with you before starting the project.
We do not replace your support team
The chatbot handles routine questions and gathers information. Your team handles cases that need their judgement. If the aim is to cut headcount, this is not the right project for us.

The work your team still needs to do

Some support work moves to the chatbot. Your team still handles the cases it passes on and takes on new tasks: reviewing conversations and keeping its information current. We agree who will do this work and allow time for it in the project.

Reviewing the answers

At launch, someone needs to review conversations, identify errors and correct the chatbot's information or behaviour. That takes time and a named owner, both of which need to be accounted for in the project.

Handling cases passed to your team

Some conversations will still need a support agent. Their job is to assess the case and resolve it. You also need to check whether the information gathered by the chatbot actually helps them do that.

Keeping information up to date

When prices, procedures or opening hours change, the source the chatbot uses needs updating. Someone must remain responsible for that work after launch.

What to measure

How many requests does your support team receive, and how many cover the same topics?
How long does the customer wait for a reply that helps them move forward?
How often does a support agent have to ask for information the customer has already provided?
How long does it take a support agent to understand the case after the chatbot hands it over?
How many hours a month are needed to review answers and keep the information current?

Facts and sources

Article 50 of the AI Act has applied since 2 August 2026. Providers must ensure that systems intended to interact directly with people are designed and developed so that people are informed they are interacting with AI, unless this is obvious to a reasonably well-informed, observant and circumspect person in the context of use. This also applies to the chatbot on your website.

Commissione europea — timeline di attuazione dell'AI Actconsulted on 2026-09-17

Questions we often hear

How long does it take to launch?

It depends on what the chatbot needs to do and which systems it needs to access. We set the timeline after reviewing the tasks, systems and connections with you.

Where is your customers' data processed?

On our website, all model processing runs on AWS Bedrock in the European Union. For your project, we agree where data will be processed and specify this in the contract.

Can it replace your switchboard or support team?

No. It can handle routine questions and gather the information your team needs to take over. People still handle cases that need their judgement and involvement.

How are answers checked for accuracy?

Conversations are available for review. We plan systematic reviews at launch, followed by spot checks. We agree with you who checks the answers and who makes corrections when an error is found.

Do customers need to know they are talking to AI?

Yes, as a rule. Since 2 August 2026, Article 50 of the AI Act has required providers to ensure that people interacting directly with an AI system are informed of this, unless it is obvious from the context of use to a reasonably well-informed, observant and circumspect person. That includes the chatbot on your website.

Does it work on WhatsApp?

Yes. We can connect it to WhatsApp and the systems needed to handle requests. We also define what it can do on that channel and how it hands conversations over to your team.

What if customers do not use it?

That can happen if it does not meet a real need. We start by reviewing the requests you already receive with you, so we can identify where a chatbot would help before building it.

Tell us what your customers ask you

We start with a few recent support requests. We will go through them with you to see whether a chatbot would help, what it should handle and what work your team would still need to do.