AI Service Guide

1. Introduction & Key Advantages

Last Update: 22nd Sep 2026

Welcome to 360.my AI Service! Our platform empowers your business to build, customize, and deploy intelligent, enterprise-grade AI Agent Chatbots directly onto customer communication channels such as official WhatsApp Business.

Whether you need an automated 24/7 customer service representative, an interactive sales assistant, or an intelligent query resolution bot, our AI suite provides end-to-end tooling from prompt engineering to automated testing and backend data integration.

Why Choose 360.my AI Service?

  • Official WhatsApp Business Integration: Connect directly with Meta-authorized WhatsApp Business numbers to engage customers on the world's most popular messaging app.
  • Multi-Tier LLM Architecture: Choose between Lite, Standard, and Professional language models to optimize for response speed, reasoning depth, and cost.
  • Domain Knowledge Base (RAG & Vector DB): Upload your own documentation, product catalogs, and FAQs. Our vector database delivers precise, grounded, hallucination-free answers.
  • Backend System Integration (MCP Servers & Tools): Equip your AI agent with Model Context Protocol (MCP) servers and tools to query live order statuses, account balances, and internal APIs in real time.
  • Automated Prompt Regression Testing: Run automated test suites to ensure prompt or knowledge modifications never break your desired customer response behavior.
  • Smart Delay Response Aggregation: Intelligently combines multiple rapid consecutive customer messages into a single thoughtful answer, eliminating fragmented replies.
  • Human-in-the-Loop Escalation: Seamlessly hand over conversations from AI to human operators inside a unified inbox whenever personal assistance is required.

2. Video Tutorial: Setting Up Your First AI Agent

This comprehensive video tutorial walks you through setting up, configuring, and deploying an autonomous AI Agent chatbot from scratch on 360.my.

Quickstart Step-by-Step Guide:
  1. Sign In or Register: Log in to your account. Newly registered accounts with completed profiling receive RM5 free testing credits.
  2. Configure WhatsApp Sender: Ensure you have an active WhatsApp Business number linked with Meta under your account.
  3. Design in AI Playground: Navigate to AI Service > AI Playground to generate and test your system instructions.
  4. Create Your AI Agent: Go to AI Agents, set your agent name, assign the channel, and configure response parameters.
  5. Add Knowledge Base: Under Agent Knowledge, input your business FAQs and manuals for semantic vector search.
  6. Run Automated Tests: Build test assertions in Prompt Tests to verify your agent passes quality criteria.
  7. Monitor & Manage: Track live interactions and manage human escalation via the unified chat Inbox.

3. AI Playground & Model Tiers

The AI Playground is your interactive testing sandbox where you can prototype, evaluate, and fine-tune your agent's persona before deploying to production.

  • Automated Prompt Generation: The system automatically generates a tailored initial system prompt based on your registered business profile.
  • Real-Time Simulation: Chat with your agent in a simulated environment to review answers and test session memory retention.
  • Flexible Model Selection:
    • Lite LLM: Lowest operational cost (hosted on dedicated RTX 3090 AI server hardware). Because response speeds are slower, it is ideal for initial testing, staging experiments, and non-urgent asynchronous tasks (such as email handling) rather than urgent live chat or real-time WhatsApp messaging.
    • Standard LLM: Balanced performance and reasoning capability; great for general customer service and product recommendations.
    • Professional LLM: Advanced reasoning and multi-step logic; best for complex troubleshooting, detailed data interpretation, and multi-tool execution.
  • One-Click Export: Once satisfied with your prompt in the playground, export it directly into your active AI Agent.

4. Agent Knowledge & Vector Database (RAG)

Empower your AI Agent with domain-specific knowledge using our Retrieval-Augmented Generation (RAG) engine powered by an automated vector database.

  • Custom Knowledge Ingestion: Add customer service manuals, product pricing, return policies, and location info directly into the knowledge base.
  • Automatic Vectorization: Our platform converts your textual documents into vector embeddings for fast and precise semantic retrieval during live conversations.
  • Built-in Version Control: Save your knowledge updates with keyboard shortcuts (Ctrl + S on Windows / Cmd + S on Mac). Every change requires a commit message, enabling transparent version diffing and instant rollback.

5. MCP Servers & Custom Agent Tools

Take your AI chatbot beyond static text replies by connecting it directly to your business databases and external services.

  • Model Context Protocol (MCP) Servers: Integrate with open standard MCP servers to let your AI query account balances, check order statuses, retrieve tracking numbers, and query live database records.
  • Agent Tools: Define customized system functions that the AI can dynamically invoke during a conversation to perform real-time actions autonomously.
  • Secure & Controlled Access: Safely expose only approved endpoints and parameters to the AI model, ensuring enterprise-grade data security.

6. Prompt Tests & Automated Regression Reports

Ensure consistent, high-quality responses every time you modify your system prompt or knowledge base with automated testing.

  • Automated Test Cases: Create test prompts mimicking actual customer questions (e.g., service introductions, delivery status inquiries, out-of-scope questions) paired with expected keyword or sentiment assertions.
  • Regression Prevention: Avoid tedious manual testing. Run your entire suite of prompt tests with a single click to instantly verify that recent edits have not caused unintended behavior.
  • Prompt Test Reports: Review detailed execution logs with clear Pass/Fail badges, response comparisons, and performance metrics.

7. Smart Delay & Human Handoff

Deliver an authentic, human-like chat experience while maintaining human oversight when complex issues arise.

  • Smart Delay Response (5–10s Buffer): Real customers often send several short fragmented messages consecutively ("Hi", "Are you available?", "Where are you located?"). Our configurable delay buffer gathers all incoming messages within the window and replies with one consolidated answer.
  • Unified Live Inbox: Monitor all incoming WhatsApp conversations in real time through the platform's unified message inbox.
  • Human Agent Takeover: Assign any conversation to a human support agent at any time to immediately suspend automated AI replies.
  • Seamless AI Re-engagement: Once the human operator resolves the issue, set the ticket back to pending to allow the AI agent to resume automated handling.
  • Custom Escalation Procedures: Specify escalation criteria directly inside your system prompt so that requests requiring human authority are systematically routed to your team.