Skip to content

Case study · Built for Kout Food Group, Kuwait

KoutGPT

Every question, answered.

An AI agent we built for Kout Food Group. It connects the company's data to a conversational interface, so authorised people can ask their business anything and get answers in plain language.

KoutGPT sign-in page: Every question, answered. A username and password form beside the line Ask your business anything
KoutGPT chat page: a Welcome back greeting above a message box that reads Reply to Database-Bot, with a microphone and send button
KoutGPT sidebar: New Chat, Projects, a Database-Bot data-source selector and a Chats and analysis list of recent sales questions

The need

Answers that live in a database.

A food-service group like Kout Food Group runs many brands. Questions about sales and performance depend on data held in databases that few people can query directly.

KoutGPT gives authorised people a direct way to ask: type or say the question as you would to a colleague, and read the answer.

The solution

Three screens, one conversation.

  1. KoutGPT sign-in page: Every question, answered. A username and password form beside the line Ask your business anything

    01

    Sign in

    Every person has their own account. The sign-in page asks for a username and password, then takes them straight to KoutGPT.

  2. KoutGPT chat page: a Welcome back greeting above a message box that reads Reply to Database-Bot, with a microphone and send button

    02

    Ask your business anything

    Questions go in as plain language, typed or spoken, in a single message box that answers from the Database-Bot data source.

  3. KoutGPT sidebar: New Chat, Projects, a Database-Bot data-source selector and a Chats and analysis list of recent sales questions

    03

    Pick up where you left off

    The sidebar keeps past conversations under Chats and analysis, with a New Chat button and a Projects area. A selector chooses which data source the questions go to.

Screens from the live app.

What it does

Simple on the surface, connected underneath.

  • Per-user sign-in

    Accounts with a username and password, and a welcome that greets the signed-in user by name.

  • Choose the data source

    A selector, shown as Database-Bot, sets what each question is asked of.

  • Type or speak

    The message box takes typed questions and has a microphone button for voice.

  • Graphs as well as text

    Answers can come back with charts as well as written figures, so a trend is seen at a glance.

  • Chats and analysis

    Past conversations are saved in a list, with search and a New Chat button.

  • Projects

    A Projects area alongside the chat list for organising work.

  • In plain language

    Ask your business anything: questions go in as you would say them, not as queries.

What we delivered

How it is put together

Built for
Kout Food Group, by BCT.
Data
A company database, chosen from a data-source selector.
Interface
A web app with per-user sign-in, typed and voice input, and saved chats.
Answers
Questions go to the selected source and the answer comes back in the conversation.

The value

Questions asked where the data lives

Authorised colleagues can put their business questions directly, in their own words, and keep each conversation for later. There is no query language to learn and no queue for someone else to run the numbers.

It shows what an AI agent can be when it is connected to real company data: not a script of canned replies, but an interface to the business itself.