Website chat
A chat bubble on your website. Customers type a question and get an instant answer.
On your website, in the browser and on the phone, in your customers' own language. It answers from your documents in about a second, and hands over to your team when it should.
Any company, in any industry (clinic, real estate, online shop, school, agency, software) signs up, uploads its information, and gets an AI assistant that works on four channels.
A chat bubble on your website. Customers type a question and get an instant answer.
A “Talk to us” button. Customers speak with the assistant in the browser. No phone needed.
Customers ring your phone number. The assistant picks up, day or night.
A customer fills an enquiry form, and the assistant calls them within minutes to help.
We run the platform. Companies set up their assistant. Their customers talk to it. The assistant does the work in the middle and reports back.
Keep scrolling to move through the eight steps. The line shows how one customer conversation travels from the company's setup to the assistant and back to the company.
No custom code per company. The industry template fills in sensible defaults, and the company adjusts them. Scroll (or click a step) to see a sketch of each screen in the company console.
The assistant only calls people who asked to hear from you. They arrive from your spreadsheets, your Meta ads, your website and your CRM. Every lead is checked first, then called in their own language.
| Column | Needed? | Example |
|---|---|---|
| Name | Required | Alex Martin |
| Phone | Required | +44 7700 900123 |
| Consent + date | To be called | yes · 2026-11-02 |
| Language | Optional | English |
| Optional | alex@example.com | |
| What they need | Optional | Knee pain |
Your column names don't have to match: you pick which column is which during the import.
Download a sample filePick a channel. The sketch on the left plays a real-feeling conversation; the steps on the right light up as each part happens.
On a call, the assistant runs this loop on every turn. Everything happens in a stream, so it starts speaking before the whole answer is even written.
The assistant detects the language, replies in it, and can switch mid-conversation. Your documents can be in any language; no translation work for you.
Example languages. Each language's voice quality tier is confirmed by test calls before launch (see language tiers below).
| Setting | How it works |
|---|---|
| Languages | Switch on the languages you want and choose a default. |
| Voices | One voice per language, or one multilingual voice so “Maya” sounds the same everywhere. |
| Greetings & disclosures | Auto-translated drafts that your team reviews before go-live. |
| Forms | Shown in every enabled language; the form language becomes the customer's language. |
| Phone numbers | One number that detects the language, or one number per language. |
| Handoff | Send transfers to staff who speak the caller's language. |
| Review | Read every transcript in the original or translated into your language. |
Voice quality differs by language, so every language gets a tier. Companies always know what to expect.
| Tier | Chat | Voice calls |
|---|---|---|
| Tier 1 | ✅ | ✅ Full quality, ~1 second replies |
| Tier 2 | ✅ | ✅ Marked “beta”, slightly slower |
| Tier 3 | ✅ | Not yet: offers chat link or callback |
A language moves up a tier only after it passes our test calls for accuracy, natural voice and speed.
The moment a chat or call ends, the platform does the follow-up work automatically.
Alex called about knee pain treatment. Asked about Saturday hours and the consultation price. Booked a consultation with Dr. Rao for Saturday 10:30. Prefers SMS reminders.
Sample data for illustration.
A working sketch of the company console with sample data. Use the menu on the left: every list opens a detail view, and every screen has a short “What am I looking at?” note.
Every call starts with “I'm an AI assistant and this call may be recorded”, in the call's language.
The AI only calls people who ticked “you may call me”. Do-not-call lists are respected. Calls only in allowed hours.
Every company's data is fully separated. One company can never see another's documents, calls or customers.
The assistant uses your documents and rules. If it doesn't know, it says so and offers a person.
Recordings are encrypted, viewable only by your authorised staff, and deleted on your schedule or on request.
Disclosures and consent wording are never machine-translated without your team approving them.
Each phase ends with something you can see and test. The order matters more than dates; timing depends on team size.
Each one has our suggested answer. Hover or tap a card to see it; you can accept it or change it.
One company using the platform. Each tenant's data is kept completely separate.
A ready-made starting setup for a type of business (clinic, real estate…), with typical questions and goals.
The documents, FAQs and website pages the assistant uses to answer.
What the assistant should get done in a conversation, e.g. book an appointment, and which details to collect.
The assistant's name and personality, e.g. “Maya from ABC Clinic, warm and professional”.
Inbound: the customer calls you. Outbound: the assistant calls the customer (after a form).
Passing the call to a real person. “Warm” means the staff member first gets a short summary.
When the customer talks over the assistant. It stops and listens, like a person would.
The pause between the customer finishing and the assistant replying. Our target is about one second.
Turning the caller's voice into text, and turning the assistant's reply back into a natural voice.
A potential customer whose details the assistant collected (name, phone, need…).
Automatically sending a lead or summary into the company's other systems.
On outbound calls, recognising an answering machine and leaving a short message instead.
The customer's permission to be called, captured on the form with time and wording.
Uploading a spreadsheet of people who asked to be contacted. You match its columns once; every row is then checked before any call.
Where a lead came from: a spreadsheet, a Meta ad, a website form, your CRM, or a chat or call.
A test area in the console where the company can chat and talk to its assistant before going live.
The complete guide covers every flow with diagrams, screen sketches and the decisions list. Read it online or download the PDF to share with your team.
Questions? The DevsTree team is at info@devstree.com.