All case studies and examples
    Flipside implementation case study

    Action chatbots

    How we built Flip, our website AI assistant.

    A practical case study of Flipside’s own assistant: approved business answers, a separate enquiry form and a clear route to booking a discovery call.

    Best suited toFlipside’s own website and Always On demo
    Action chatbotsConnected system
    Curated knowledge
    Separate enquiry form
    Booking route
    Clear limits

    Our own website implementation. Describes the delivered functionality; no conversion uplift or client result is claimed.

    The brief and the boundaries

    Flipside offers consulting, automation, voice agents and website assistants. A visitor may arrive with a task rather than a product name. We built Flip to explain those services using maintained business information, then direct the visitor to a relevant page or a human conversation. Always On has a separate demo focused on the website-assistant package.

    Keep the knowledge deliberate

    Flip receives curated facts about Flipside, its services and process. The Always On demo has its own offer-specific knowledge, including pricing and exclusions. Neither assistant automatically crawls a visitor’s website or learns new business policy from a chat.

    Separate conversation from submission

    Chat explains the offer and points to the next step. A separate form collects name, email and phone for an enquiry. The chat cannot silently send an email, create a CRM record or confirm a booking. Those actions need their own integration and checks.

    Make failure visible

    When live chat is unavailable, the interface distinguishes sample replies from live answers. Failed requests display an error. The enquiry form shows success only after the sending service accepts the request; that is not a guarantee of inbox delivery.

    Measure the real next step

    Opening the assistant, starting a conversation, submitting an enquiry and completing an embedded discovery booking are separate events. The implementation keeps chat text and form values out of these analytics events. A booking-page visit is not counted as a completed appointment.

    A worked scenario

    An illustrative walkthrough of the workflow, not a transcript of a customer conversation.

    Visitor question
    Can the website assistant book appointments?
    Answer boundary
    The standard package links to an existing booking page. Live availability and confirmed bookings require a separately scoped integration.
    Next step
    The visitor can inspect package pricing, submit the enquiry form or choose a discovery-call time.
    Human responsibility
    Lucas reviews the business’s actual tools and scope before promising a specific integration.

    What was built, and what still needs measuring

    The implementation provides two distinct assistant experiences, package information, an enquiry path and an embedded booking journey. You can inspect the public demo and booking page below. Its effect on lead quality, time saved and conversion rate still needs a measured baseline and verified analytics collection.

    The challenge

    The friction this system is designed to remove.

    01Visitors need to distinguish consulting, custom builds and the standard website-assistant package.

    02The assistant needs a maintained source of business facts and an explicit boundary around what it can promise.

    03A chat message, a submitted enquiry and a confirmed booking must remain distinct steps.

    The system

    Four connected parts, one useful outcome.

    01

    Curated knowledge

    Use separate maintained facts for the main website assistant and the Always On demo.

    02

    Separate enquiry form

    Collect name, email and phone only when the visitor chooses to enquire.

    03

    Booking route

    Link to the discovery page, where Calendly handles appointment selection and confirmation.

    04

    Clear limits

    Chat does not submit enquiries or make bookings. Form values and chat text stay out of analytics events.

    Example workflow

    From request to result.

    01

    Question

    A visitor explains what they need using their own words.

    02

    Context

    The chatbot finds the relevant approved information and identifies the next question.

    03

    Choice

    The visitor chooses to explore a service, enquire or open the booking page.

    04

    Completion

    The separate form or booking integration confirms whether the requested step succeeded.

    Designed with control

    Safeguards built into the workflow.

    Clearly identified as AI

    Approved answer boundaries

    Privacy-aware data collection

    Human handover

    Measurement plan

    What to check against a baseline.

    Conversation completion

    Qualified enquiry rate

    Booking conversion

    Human correction and escalation rates

    Explore further

    AI agents and chatbots compared

    Read the insight
    Related service

    See how Flipside approaches this build.

    Explore the service
    Your workflow

    What would this look like
    inside your business?

    Bring us the current process and the result you want. We will map the smallest useful version.

    Book a free 15-minute discovery call