Overcome the limits of generic LLMs by combining them with your internal knowledge to deliver fast, accurate answers.
Speed up customer support by referencing your support manuals; no rule-based scripting required.

Why now?

Hardware progress and model advances (e.g., Transformers) have made AI practically useful. Adopting chatbots earlier improves productivity.

What to do next

Press Start below!

2 - Getting Started

Start using the DfocusGPT chatbot

Read this first to get set up quickly.

This section helps you try the project yourself.

Prerequisites

Prepare the support manuals that the chatbot should reference.
Sign in with an email account. Assign at least one admin.

Case studies

Dfocus website intro chatbot – answers questions about our site and blog.
iQuest ERP chatbot – consults on ERP; supports agent handoff.

Initial setup

Agent management – register agents and assign categories.
Knowledge management – upload Q&A or documents (Excel/Word bulk upload supported).

Demo

Request a demo: gptsupport@dfocus.net

2.1 - Case Studies

How DfocusGPT is used in real sites.

The customer information below is posted with prior consent. Unauthorized copying is strictly prohibited.

Dfocus Website Chatbot

Dfocus delivers projects with global solutions and in‑house tools. Explaining years of accumulated solution expertise is hard even for internal experts. The Dfocus website chatbot provides natural‑language answers about our solutions using DfocusGPT.

alt text

*Even topics like technical log in SeedVDI are explained clearly via LLM answers.

iQuest ERP Support Bot

iQuest provides a reasonably priced ERP for SMEs, including the “얼마에요 ERP”, enhanced with AI automation. With tens of thousands of active users and heavy site traffic, agents must guide many users across various solutions. DfocusGPT provides natural‑language initial responses and simple answers to reduce agent load.

*The bot clearly answers that 얼마에요 ERP 4.0 is a SaaS cloud solution usable immediately after signup without separate hardware or installation.

Notes

Warning

Copying customer information is strictly prohibited.

Contact

gptsupport@dfocus.net

3 - Answer Examples

See how the chatbot answers.

Answers vary based on the question and reference materials.

Guided steps – provides step-by-step instructions when menus are unclear.
Consolidated answers – gathers scattered info for concise summaries.

4 - Core Concepts

What your customers expect and what to deliver.

Narrow the knowledge gap – make organizational knowledge instantly accessible.
Instant service – deliver answers quickly via an AI chatbot.
Keep it intuitive – provide only what’s needed at a fair price.

Below is a basic sequence of how user questions are answered.

sequenceDiagram
    autonumber
    loop History
    User->>Chatbot: Question
    Chatbot->>Company KB: Search
    Company KB->>Chatbot: Results
    loop LLM
    Chatbot->>LLM: Prompt (Q + KB + History)
    LLM->>Chatbot: Answer
    end
    Chatbot->>User: LLM answer + sources
    Note right of Chatbot: Store Q&A
    end

Flows may vary by options. Agent handoff and KB management flows are similar.

5 - Key Features

Knowledge management, chatbot, and agent handoff.

We update docs continuously; check the service for the latest UI.

We start with the main features.

5.1 - Home Screen

The initial screen after login.

Tooltips on each icon describe the menu items.

alt text

Logo – returns to the home screen.
Calls – pending, active, and finished agent calls.
Notifications – async task notifications (e.g., upload complete).
User – email menu with password change, logout, category switch.
Status – set agent status (busy, offline, etc.).
Agent management – add agents, set nicknames/brands.
Consulting – view user info and chat history; respond appropriately.
Chat history – user, bot, and agent conversations.
Internal messenger – chat between signed‑in agents.
Q&A Management – manage sources for chatbot answers.
Recommended QA - from bulk text typed(or copied or upload), generate some recommedded FAQ and add it into Q&A database.
Analytics – dashboards for load, sessions by agent/category, usage details.
Settings – default phrases, working hours, and embed scripts.
Chatbot testing – test performance after entering knowledge.
Inquiries – view your submitted requests.
Contact – submit quick questions while using the system.

Intuitive admin UI

Bulk upload searchable Word documents with images; edit entries easily.
Build a dedicated “manual‑style” chatbot window that renders images nicely.

alt text

5.2 - Chatbot Testing

After entering Q&A data, test answer quality via chat.

After organizing knowledge, test the chatbot with various questions.

Left panel shows “New chat” and chat history.
Type your question and press Enter or click the paper‑plane icon to start.
Q&A is saved to history automatically.

Click “More info” under an answer to see the sources referenced by the chatbot. alt text

5.3 - Q&A Management

Manage question–answer pairs used for RAG search.

Manage frequently asked questions and their answers.

alt text

Add – create an FAQ.
Edit – modify an FAQ.
Delete – remove an FAQ.
Upload documents – see Bulk Upload.
Search – keyword/Semantic matching.

Intuitive and convenient Q&A management

Bulk upload and edit Word documents with images.
Generate readable answers like “manuals with images”.
Streaming answers minimize perceived latency.

5.4 - Agent Console

Agents respond to “Connect to agent” requests here.

When a user requests agent handoff, an agent connects automatically or manually and continues the chat.

The user selects “Connect to agent”.
The agent starts the session from the waiting list.
The middle panel shows the prior user–bot conversation.
The user sees “Agent connected”.
The agent chats with the user.

alt text

sequenceDiagram
    autonumber
    loop History
    User->>Chatbot: Question
    Chatbot->>Company KB: Search
    Company KB->>Chatbot: Results
    loop LLM
    Chatbot->>LLM: Prompt (Q + KB + History)
    LLM->>Chatbot: Answer
    end
    Chatbot->>User: LLM answer + sources
    Note right of Chatbot: Store Q&A
    end
    User->>Chatbot: Request agent handoff
    Chatbot->>User: Ask phone/name
    User->>Chatbot: Provide phone/name
    Chatbot->>Agent: Waiting for connection
    Agent->>Chatbot: Start
    Chatbot->>Agent: Provide chat history
    Agent->>User: Chat or phone
    Note right of Chatbot: Store Q&A

Video demo (Korean):

5.5 - Bulk Upload

Upload Excel and Word files in bulk.

Bulk uploads must follow the required formats.

Excel files must be Question/Answer pairs (Q in the first column, A in the second).
Word files also follow Q/A pairs. See the Word section.

After preparing the file, click “Upload Document” and select your file. alt text

5.5.1 - Excel File

Excel format for bulk upload.

Follow the template below.

Excel template

Save an Excel file in the following format and upload it to import Q&A entries. The first row is a header and is ignored.

Question	Answer
Q1	A1
Q2	A2
Q3	A3

5.5.2 - Word File

Word format for bulk upload.

Use the provided Word sample format.

When uploading a Word file with images, images are stored along with text.
Like Excel, structure the document as Q/A pairs. If the document can clearly separate Question/Answer (e.g., using labels), it can be bulk‑uploaded.
After upload, you can edit entries in Q&A Management.

alt text