Ganchev Solutions
AI Solutions Studio

A practical workflow example

Customer enquiry automation,
with a person in control.

An enquiry comes in. A draft is prepared using approved information. Your team checks the answer and decides what happens next.

Interactive simulation. These are prepared examples for a fictional meeting-room business, not Ganchev Solutions services or client results. No live AI model, inbox or booking system is connected. Nothing is sent or reserved.

  1. 01

    Read the enquiry

    Identify the request and any missing information.

  2. 02

    Prepare a draft

    Use the business facts. Leave unknowns open.

  3. 03

    A person decides

    Check the reply or take over an exception.

Try the example

Same workflow. Three different enquiries.

Choose a prepared scenario, compare the draft with the reference facts, then try the review step.

01 · Incoming enquiry

A routine question

We are a team of six. Does your meeting room have Wi-Fi and a screen for a presentation?

02 · Prepared example output

Prepare an information reply

The question can be answered from the approved room and equipment information. No booking is being requested or confirmed.

Reference: Room · Equipment

Draft reply for a person to check

Hello, our meeting room accommodates up to eight people, so a group of six is within its stated capacity. Wi-Fi and a presentation screen are included. If you would like to enquire about a booking, please share your preferred date and time so our team can check availability.

03 · Human review

Would you use this draft?

Check all three before marking the sample ready.

Review progress: 0 of 3 checks complete. Nothing has been sent.

From example to implementation

What a real workflow needs.

This page shows the decisions around an AI-assisted draft. A live implementation needs its own business information, connections and testing.

  1. Agree the job and its limits.

    Choose which enquiries the workflow can help with, what a useful draft looks like, and which requests always go to a person.

  2. Use maintained reference information.

    Connect the approved source for services, terms and common questions. Decide who updates it and what happens when information is missing or out of date.

  3. Connect tools with defined permissions.

    A real model could prepare the draft from an enquiry and that reference. Agree how the inbox or customer system is accessed, which data is used, and where a person reviews the result.

  4. Test before relying on it.

    Test routine questions, ambiguity, complaints and attempts to override instructions. Review incorrect answers and measure time spent checking drafts as well as time spent writing them.

Read the three prepared examples.

Compare the enquiry, the prepared response and what a person needs to check. These examples can also be read with JavaScript turned off.

A routine question

Customer enquiry

We are a team of six. Does your meeting room have Wi-Fi and a screen for a presentation?

Prepare an information reply

The question can be answered from the approved room and equipment information. No booking is being requested or confirmed.

Reference: Room · Equipment

Prepared draft reply

Hello, our meeting room accommodates up to eight people, so a group of six is within its stated capacity. Wi-Fi and a presentation screen are included. If you would like to enquire about a booking, please share your preferred date and time so our team can check availability.

Review before using the draft

  • The capacity and equipment match the reference facts.
  • The reply makes no promise about availability or a confirmed booking.
  • It answers the customer’s question clearly.
Some details are missing

Customer enquiry

Hello, we need a room next week. Can you send me a price and reserve it for us?

Ask for the missing details

The date, time and group size are missing. A useful draft asks for them without inventing a price or claiming to reserve a room.

Reference: Booking hours · Confirmation

Prepared draft reply

Hello, thank you for your enquiry. Could you share your preferred date, start and end times, and the number of people attending? Our booking hours are Monday to Friday, between 9am and 6pm. With those details, our team can check availability and prepare a quote. A booking has not been reserved at this stage.

Review before using the draft

  • The reply asks for the date, times and group size.
  • The opening hours match the reference facts.
  • No price, availability or reservation has been invented.
A person needs to decide

Customer enquiry

We now need the room for 12 people on Thursday. Please confirm you can fit everyone, and refund half of what we paid for our last booking.

Hold for the manager

The group exceeds the stated capacity and the message asks for a refund. The supplied information cannot authorise either request.

Reference: Room · Exceptions · Confirmation

Prepared internal handoff note

Internal handoff note: the customer asks for capacity for 12 people and a partial refund for a previous booking. The room’s stated limit is eight. Ask the manager to review the capacity request and check the earlier booking and payment records before deciding how to respond. No booking change or refund has been confirmed.

Manager review required

  • Verify the earlier booking and payment records.
  • Check what can actually be offered for a group of 12.
  • Have the manager decide on the refund and write or approve the customer reply.

Built around your work

Which step would help your team?

This is one way AI business process automation can support a recurring task. AI Solutions Studio scopes custom workflows; the One-Workflow Implementation Sprint focuses on one agreed workflow or assistant. If you are still choosing where to begin, explore the AI Opportunity Map example.

Discuss your workflow