Human-centered design systems and product experiences
Core services
01UI/UX Design02Web Development03Software Development042D/3D Experiences05App Development06AI Agents & Automation07Website DesignTechnical services
Frontend Engineering
React DevelopmentNext.js DevelopmentWeb ApplicationsEnterprise DashboardsSaaS PlatformsBackend & Infrastructure
API DevelopmentCloud InfrastructureDevOps & CI/CDBackend SystemsCustom SoftwareIndustries we serve
Technology & Software
Website AI Chat Agent Development
Visitors answered while they browse
An agent for the chat window on your own website.
An agent for the chat window on your own website. It answers visitors from knowledge you approved, tells a browser from a buyer with your own questions, takes a booking or an enquiry through to a record, and hands over to your team with the conversation attached.
Website chat fails in two ways. Either it is a form with a chat bubble, collecting an email address and answering nothing, or it is an eager assistant that answers everything, including questions your business has no approved answer to. Visitors learn quickly which one they are talking to. The useful version sits in between: it answers what you have decided, and it is honest about the rest.
Parts of this workflow run today in our demonstrations on test data. The conversation logic behind it runs today in our demonstration console as a web chat on test data. What is not built is the chat window for your site, visitor identification and the hosted service behind it. The rest of what this page describes is a proposal that we build with you.
How the Work Is Done Today
A bubble that answers nothing
The visitor asks a real question and is told to leave an email address.
Answers nobody approved
A fluent reply about pricing or policy that your team never agreed.
No difference between visitors
A student researching and a buyer ready to book get the same treatment.
Anonymous by default
The visitor is a session, not a person, until they choose to say who they are.
Chats that end nowhere
The conversation closes and nothing reaches the CRM or the calendar.
Handover into silence
The visitor is told someone will be in touch, outside working hours, with no time given.
Six jobs it carries through
Approved answers
Answers from the knowledge you approved.
- Finds the approved answer and its source
- Links to the page that explains more
- Says when there is no approved answer
- Warns when a source is out of date
Visitor qualification
Finds out what the visitor needs.
- Asks your questions one at a time
- Applies criteria you can read
- Keeps the visitor's own words
- Stops when it has enough
Booking
Books a call or a visit against your real calendar.
- Offers times that are actually free
- States date, time and time zone
- Confirms only when the calendar accepts
- Handles a change or cancellation
Enquiry record
Writes the enquiry where your team works.
- Creates or updates one contact
- Records the page the chat started on
- Assigns an owner
- Reads the record back before confirming
Page awareness
Uses the page the visitor is on.
- Knows which service page is open
- Answers about that service first
- Suggests the next relevant page
- Does not track across other sites
Handover with context
Passes the chat to a person.
- Live when your team is online
- As a ticket when it is not
- With the conversation and the answers given
- With an honest reply time
One Case, Start to Finish
An example of one case, in 6 steps. The same steps play in the picture at the top of this page.
Chat opened
A visitor on your pricing page asks whether a service is available in their country.
Question answered
The reply comes from your approved information and names where it is written.
Need established
The visitor says they want a quote. Your qualification questions follow, one at a time.
Time booked
A discovery call is offered from your real calendar and confirmed with its time zone.
Record written
The contact, the answers and the page the chat began on are saved in your CRM.
Team informed
Your team sees the booking and the conversation, and can join the chat if it is still open.
Walk One Real Case Through It
Thirty minutes, one case from your own process. We map it against the workflow on this page and tell you plainly what can be reused from our demonstrations and what has to be built for your systems.
What It Reads and Writes
The systems named here are the kind this agent works with. Which ones it connects to in your case is settled in the scope.
Website
- A chat window built for your site, in your design
Knowledge
- Your site content
- help centre
- approved policies
Calendar
- Calendly
- Google Calendar
- Microsoft 365
- or your booking system
CRM
- HubSpot
- Zoho
- Salesforce
- or a custom CRM
Handover
- Live chat tools
- help desk
- a team inbox
Hosting
- A service that holds conversations and credentials, separate from your website
Honest About What It Knows and Who It Is Talking To
A website visitor is anonymous. The agent says so in how it behaves: it shares general, approved information freely and asks for proof before anything tied to an account. When no approved answer exists, it says that and offers a person.
The chat window is part of your website. The agent behind it is a separate service, because customer conversations and credentials do not belong in a public page's code.
- Answers only from approved knowledge
- Visitors anonymous until they say otherwise
- Contact details only with consent
- Bookings confirmed by the calendar
- Honest reply times at handover
- A record of every conversation
What Buyers Ask First
How is this different from a standard chatbot plugin?
A plugin answers from whatever it was given, in its own window, with its own data store. This is built around your systems and rules: your approved knowledge, your calendar, your CRM, your qualification criteria and your handover. It is custom development, not a subscription.
How is this different from the WhatsApp agent?
The logic is shared. The situation is not. A WhatsApp sender has a phone number and wrote to you on purpose. A website visitor has no identity and may be on any page, so this agent relies on the page context and asks before it stores anything.
Does it work on our existing website?
Yes. The chat window is added to your site and designed to match it. The conversations run on a separate hosted service that we set up with you, connected to your calendar and CRM in test accounts first.
What exists today?
Some of it. The conversation logic behind it runs today in our demonstration console as a web chat on test data. What is not built is the chat window for your site, visitor identification and the hosted service behind it. The remaining steps on this page are a proposed workflow. We would build them with you and prove each one on test data before it touches a live system.
Start With the Questions Your Site Already Gets
Bring your contact-form enquiries from the last quarter. They show what visitors ask, and which answers you have already approved.
