AI services for your website

An AI chatbot trained on your own content

A generic chatbot that answers whatever it likes does more harm than good. We install an assistant trained on your own material: FAQs, product information, service descriptions, delivery and return terms, price lists and documents. It answers from what is actually on your site, not from the internet in general.

Two things are mandatory. First, handover to a human: if the bot does not know, or the customer asks for a person, the conversation is handed over with its full history. Second, lead capture: when a conversation shows real intent, the assistant asks for contact details and the enquiry reaches you by e-mail instead of disappearing into a chat window.

AI-powered site search

Ordinary search only finds exact words. AI-powered search finds meaning: if a customer types "something that holds a phone in the car", they find the phone mount even if that word never appears in the description. On e-shops and content-heavy sites this is the most direct way to recover lost sales, because a visitor who searches is already close to buying.

AI content assistant and translations

Writing and translating product descriptions is the work that usually never gets done, simply because there is too much of it. We set up a content assistant that drafts descriptions from your product data and translates them into other languages with a consistent tone and vocabulary.

Human review always stays in the loop. The draft comes from the machine, the decision comes from a person — so no wrong fact, wrong price or wrong promise ever reaches the public site.

Lead qualification on the contact page

If you get ten enquiries a day and two of them are real, your time goes to the wrong ones. A qualification assistant on the contact page asks a few clarifying questions — what project, what scope, what timeline — and forwards the enquiry with a summary. You see in your inbox what it is about and can answer the important ones first.

Alt texts and metadata at scale

Nobody writes a thousand alt texts by hand on a site with a thousand images. AI does that work and we review the result. The same applies to meta descriptions: every page gets its own title and description, correct in length and unique. That improves accessibility and visibility in search and AI answers at the same time.

• Alt texts for every image, genuinely descriptive and aligned with accessibility requirements.
• Unique titles and meta descriptions for every page and product.
• Structured data so search engines and AI models read the content correctly.

Data protection and GDPR principles

Adding AI to a site means somebody processes data. Our principles are simple and we put them in the agreement: your content and customer data stay under your control, conversations are not used to train models without your consent, retention periods are agreed, and the visitor can see they are talking to an AI. These are technical and contractual principles, not legal advice — a regulatory assessment comes from your lawyer.

How the rollout runs

We start from your actual bottleneck: support load, not being found in search, content volume or lost enquiries. We pick one or two solutions, configure them in staging, train them on your material, test with real questions and only then go live. After launch we review conversation logs and refine the answers — an AI solution gets better in use, not at the moment it is installed.

Ongoing monitoring and refinement usually runs alongside website maintenance, so somebody actually watches the answers. If the goal is to turn those conversations into sales, we tie it into the sales growth work.