NLENFR

Chatbot & website widget

Train your digital colleague and place the chat on your own website.

Chatbot & website widget

Chatbot widget on the website
Train and embed on your site.

The chatbot is the written side of your digital team member. The same knowledge your assistant uses on WhatsApp is also used in the chat window on your website. So you only train once. In this chapter you'll learn how to fill the chatbot with knowledge, set the tone, place the widget on your site and check that everything is correct before your customers see it.

What you need

Step 1 — Gather your knowledge

A chatbot is only as good as the information you put into it. So don't start in the dashboard, but with yourself: write down the ten to twenty questions customers ask most often. Look at your mailbox and your conversations from recent weeks — customers ask those questions in their own words, and those exact words are what they'll type into the chat window. Watch for three categories: practical questions (opening hours, service area), commercial questions (what does it cost, how quickly can you come) and threshold questions (what if it doesn't work, what about the guarantee). It's often this last group that decides whether someone gets in touch.

Step 2 — Train the chatbot

  1. Open Train bot in the dashboard.
  2. Add a question-answer pair for each topic. Keep the answer short: two to four sentences, with one clear next step.
  3. Do you already have a lot of material? Import a file with questions and answers in bulk, or let our AI suggest drafts based on your website. You always edit yourself.
  4. Test each new topic immediately in the preview window. Deliberately phrase the question slightly differently from how you entered it — that shows whether the assistant recognises the intent.
  5. Save. Changes are active immediately, including in the widget already on your site.
Write answers as you would give them on the phone and end with a concrete next step: a quote form, a call appointment or a page on your site. More background: Knowledge base — Training your bot.

Step 3 — Persona: name, tone and transparency

Knowledge determines what your assistant says, the persona determines how. You give your digital colleague a name, a role and a tone of voice: formal or informal, short or detailed, with or without emoji. You also set multilingual support here, so that a visitor in another language automatically receives an answer in their own language.

Set the persona via Train persona in Slimbot. If you work with multiple numbers or brands, each number can have its own persona with its own name.

We strongly advise having the assistant introduce itself as a digital colleague. One sentence suffices: "I'm the digital assistant of [company] — a colleague will take over as soon as necessary." That's honest and customers accept a quick digital response perfectly well, as long as it's clear that a human remains available.

Step 4 — Place the widget on your website

  1. Go to Chat widget in the dashboard (appearance and embed are also under Chatbot).
  2. Set the appearance: colour, position and the greeting visitors see first.
  3. Copy the embed script, or download a ready-made CMS package (WordPress plugin or Shopify snippet) — token and instance ID are already included.
  4. Paste the code just before </body>, or install the CMS package following the steps below.
  5. Publish, open your site in a private window and check that the chat window appears in the bottom right.

If you later change the colour or greeting in the dashboard, the visitor sees that immediately — you do not need to paste the embed again (unless you have set fixed overrides in the code).

Conversations from the widget arrive on the Website tab in your inbox, alongside WhatsApp and e-mail. No need to configure per page: the widget recognises your account and branding.

Installation per CMS

Below is the fastest route per platform. If you do not have WordPress or Shopify, the embed script from Chat widget will suffice.

WordPress

  1. Log in to Slimbot and download the WordPress package: slimbot-messenger-wordpress.zip (only when logged in).
  2. In WordPress: Plugins → Add New Plugin → Upload, upload the ZIP file and activate Slimbot Messenger.
  3. Go to Settings → Slimbot Messenger and check token and instance ID (already filled in on download).
  4. Optional: colour, display name, greeting or position. Then open your site in a private window.

Shopify

  1. Download slimbot-messenger-shopify.zip (logged in) or copy the Liquid snippet from the CMS package.
  2. Add it as a theme snippet slimbot-messenger.
  3. In layout/theme.liquid, just before </body>: {% render 'slimbot-messenger' %}.
  4. Save and test in a private window. Without ZIP: replace YOUR_TOKEN and YOUR_IID manually.

Webflow

  1. Project Settings → Custom Code → Footer Code.
  2. Paste the embed script from Chat widget.
  3. Publish and check on the live site.

Wix

  1. Settings → Custom Code (or Tracking & Analytics).
  2. New code, placement: Body — end.
  3. Paste the Slimbot embed script and publish.

Squarespace

  1. Settings → Advanced → Code Injection → Footer.
  2. Paste the embed script and save.
  3. On plans without Code Injection: use a Code block on the relevant pages.

Step 5 — Testing before you go live

Do a test round with three types of questions: one the bot should know (opening hours), one that falls just outside the knowledge (a very specific price question) and one in which you ask for a human. This way you see whether the answers are correct and whether the conversation is transferred properly when it gets difficult. Also have a colleague test: they'll ask questions you'd never have thought of yourself.

Handover to a human colleague

A good assistant knows when to stop. Provide an escape route: a phone number, a call request or a notification in the inbox. As soon as you respond yourself, the assistant hands the conversation to you. Record which topics always go to a human: complaints, damage, medical questions or non-standard pricing agreements.

Automate once the basics are in place

Measure and adjust

Review ten conversations weekly. Questions the assistant evaded or forwarded are exactly the topics missing from your knowledge base. Add them and the gap is closed. Do this for fifteen minutes a week in the first month; after that, maintenance is minimal.

Common mistakes and how to fix them

Practical examples by sector

Frequently asked questions

Do I need to train the chatbot separately for my website and for WhatsApp?
No. The knowledge base is shared: what you add once, your digital assistant uses on the website, on WhatsApp and in the other channels. Only the persona can differ per number or brand.

Can I see what the chatbot has answered?
Yes. All website conversations are fully in your inbox under the Website tab, including time and progress. You can step in at any time and take over the conversation.

Does a visitor notice they're talking to AI?
That is as it should be. We advise the assistant to introduce itself as a digital colleague and make clear that a team member can take over. That is transparent and works better in practice than pretending.

Does the widget work on WordPress, a webshop and a self-built site?
Yes. For WordPress and Shopify there are ready-made packages; elsewhere you paste one line of code before the closing body tag. Any system with a script or code field works.

What happens outside opening hours?
The chat simply keeps working. With a workflow, you can have the assistant give a different response outside office hours and, for example, collect a call-back request, which you'll find in your inbox the next morning.

Which plan includes the chatbot and full persona training?
The chat function is part of the standard channels; advanced persona training is in the higher plans. On Plans you can see exactly what is in your plan and what an upgrade adds.