Fast Line Pro Overview

High-level overview of Fast Line Pro in ConnectiveOne. This page explains what Fast Line Pro is, how it works, how to prepare a knowledge base, how to write a good agent prompt, common use cases, and how to get started. A short overview video is at the top.

Who This Page Is For

Support managers and business owners — want to scale front-line support without proportional headcount growth.
Integrators and configurators — create agents, knowledge bases, and embed AI into bot scenarios.
Operators and team leads — work in a team where Fast Line Pro handles routine requests automatically, while complex cases are passed to people.

What Is Fast Line Pro

Fast Line Pro is a ConnectiveOne module for automated AI responses based on a knowledge base. You upload your documents (FAQs, instructions, policies, product descriptions), the system breaks them into searchable fragments (chunks), and stores them for fast retrieval. When a customer asks a question, an AI agent finds the relevant fragments and formulates an answer — this is called the RAG approach (Retrieval-Augmented Generation).

Key components:

Knowledge base — a container of uploaded documents split into chunks for search.
AI agent — an intelligent assistant that uses the knowledge base and your instructions (prompt) to formulate responses.
Chatbot — a simplified version of an agent for typical Q&A scenarios.

Fast Line Pro is not a replacement for operators — it is their "fast front": routine questions are handled automatically, while complex ones go to people with the context already available.

Why Your Business Needs This

Problem	How Fast Line Pro Solves It
Operators spend time on repetitive questions	The agent answers automatically from the knowledge base
Customers wait in queue	Responses are instant, 24/7
Different operators give different answers	Single source of truth — your documents
Scaling requires hiring more staff	More requests handled without headcount growth
Knowledge lives in people's heads — it leaves when they do	Knowledge is captured in the base and available to the agent

Additional benefits:

Response quality control — you can check agent responses (e.g., whether the customer asks for an operator) and automatically escalate to a human.
Update in one place — edit a document, update the knowledge base, and the agent immediately uses the new information.
Auto-sync — set up automatic knowledge base updates from an external source (website, Notion, etc.) via KBPlugin.

How It Works — Step by Step

Prepare your knowledge base. Upload files (TXT, MD, PDF, DOCX, XLSX, and others) to Fast Line Pro. The system processes them and splits them into fragments (chunks).
Create an AI agent. Set a name, connect the knowledge base, choose an AI model, and write a prompt — the instruction that defines the agent's behavior.
Embed it in a scenario. In a bot (widget, Telegram, Viber, WhatsApp, etc.) add a step: "customer message" → call the Fast Line Pro agent → reply to customer. Optionally add response validation and escalation to an operator.
Customer asks a question. The agent searches the knowledge base for relevant fragments, formulates a response, and sends it to the chat.
If the agent is unsure or the customer asks for a human — the scenario can transfer the dialog to an operator who sees the full context and continues the conversation.

Knowledge Base: How to Prepare It Well

The quality of agent responses directly depends on the quality of the knowledge base. Here are the recommendations.

Supported File Formats

Supported: TXT, MARKDOWN (MD), MDX, PDF, HTML, XLSX, XLS, DOCX, CSV, HTM.

For best processing, TXT or Markdown is recommended.

File Structure

The best approach is a "Question–Answer" format separated by --- blocks:

Possible questions: How do I cancel an order? Can I return a product?
Answer: To cancel an order, go to your account, find "My Orders",
select the order and click "Cancel". Returns are accepted
within 14 days of receiving the product.

---

Possible questions: What is the delivery cost? How much is shipping?
Answer: Delivery is free for orders over $50.
For orders under $50, the shipping cost is $5.

You can also use a factual listing format — just blocks of information without questions.

Tips for Your Knowledge Base

Write clearly — imagine explaining to a new support agent. Avoid internal jargon without explanation.
Add phrasing variations — customers ask differently: "how to return", "refund", "I want to send it back". More variations = more accurate search.
Keep it current — outdated answers are worse than no answer. Update the base when conditions, promotions, or processes change.
One topic per block — don't mix delivery and returns in one chunk. Use --- separators.
Test after uploading — try real customer questions and fill gaps where the agent can't find an answer.
Use KBPlugin for automatic sync if your knowledge lives in an external system (website, Notion, wiki).

Agent Prompt: How to Write It Right

The prompt (instructions) is the text that defines the agent's role, style, and behavior rules. It directly affects how well agent responses match your expectations.

What to Include in a Prompt

Agent role. Who is it: "You are a support consultant for Company X", "You are a virtual assistant for online store Y".
Tone and style. Friendly, professional, formal, casual. For example: "Respond briefly, in a friendly manner, in English."
Behavior rules. What the agent should NOT do:
- "Do not make up information that is not in the knowledge base"
- "If you don't know the answer — tell the customer you will pass their question to an operator"
- "Do not discuss competitors"
- "Do not give legal advice"
Business context. Information the agent should know: company name, working hours, key policies.
Response format. For example: "Respond in 2–3 sentences, no lists" or "Structure the response as a list when there are multiple points".

Prompt Example

You are an AI support assistant for "QuizBot" company.
Your role is to help customers with questions about products, delivery, and returns.

Rules:
- Respond in English, in a friendly and professional tone.
- Base your answers strictly on the knowledge base.
- If the answer is not in the knowledge base — tell the customer you will pass
  their question to an operator.
- Do not invent or add information beyond the knowledge base.
- Be concise: answer in 2–4 sentences.

Context:
- The company operates 9 AM–6 PM (Mon–Fri).
- Free delivery for orders over $50.
- Returns accepted within 14 days.

Tips for Your Prompt

Start simple — a basic role + one rule "don't make things up". Add detail gradually.
Test and iterate — ask colleagues to send real questions. If the agent answers incorrectly, refine the prompt.
Use variables — {clientName}, {companyName}, and others for personalized responses. Variables are configured separately in agent settings.
Specify fallback behavior — "transfer to operator", "offer to leave contacts", "say the information is being clarified".
Limit topics — if the agent should only answer about delivery, state it explicitly. Otherwise it may "improvise" on topics where it lacks quality data.

Checking Agent Responses

The agent may not always give a perfect answer. ConnectiveOne lets you build validation into the scenario:

Keyword check — if the agent's response contains "transfer to operator", "I don't know", "please contact" — the scenario can automatically switch the dialog to a human.
Empty response check — if the agent could not find an answer, the customer is not left without help.
Fallback scenario — if the agent cannot handle the request, an alternative scenario is triggered (e.g., operator queue or a request for contact details).

This gives you flexibility: AI handles the main flow, while critical situations remain under control.

Start with one knowledge base for the most common questions — delivery, returns, payment, promotions.
Prepare a structured file in "Question–Answer" format or fact blocks. File quality = answer quality.
Create one agent with a simple prompt — role, tone, one rule "don't make things up".
Embed it in one channel — for example, the website widget. Test with real questions.
Analyze the results — where does the agent make mistakes? Where can't it find an answer? Fill the knowledge base gaps and refine the prompt.
Connect other channels — Telegram, Viber, WhatsApp. Add categorization or separate knowledge bases for different products.
Set up auto-sync via KBPlugin if your knowledge lives in an external system.

What You Get

Instant responses to customers 24/7 on routine questions without operator involvement.
Single source of truth — responses are based on your documents.
Easy knowledge updates — edit a document, update the base, the agent immediately uses the new information.
Flexible scenarios — combine Fast Line Pro with categorization, response validation, operator handoff, and other bot steps.
Team relief — operators focus on complex cases, escalations, and personal service.

Important to Remember

Answer quality = knowledge base quality. Documents must be clear, structured, and up to date.
AI can make mistakes. Review important cases. Always give customers a way to reach an operator.
The prompt is an investment. Spend time writing and testing it — this pays off in answer quality.
Start small. One agent, one knowledge base, one channel. Scale gradually.
Visibility and access to Fast Line Pro and knowledge bases are determined by your system's permission settings.
Fast Line Pro helps the team, it does not replace it. Operators remain for complex questions, escalations, and personal service.

Next Steps: Detailed Guides

Use these guides to go from overview to concrete actions: