> ## Documentation Index > Fetch the complete documentation index at: https://help.agentsy.pro/llms.txt > Use this file to discover all available pages before exploring further. # Understanding User Intent > Recognize what visitors want to accomplish and create more helpful, relevant AI conversations. # Understanding user intent Visitors don't all arrive with the same goal. Some want to learn, others are comparing options, and some are ready to speak with a person. Understanding **user intent** helps your Agentsy AI Experience respond appropriately without turning every interaction into a sales pitch. > [!NOTE] > > Intent is an interpretation of what someone communicates, not a guaranteed prediction of what they will do. Treat it as context for helping the visitor—not as a definitive lead score. ## Start with the visitor's question A useful first response addresses the question the visitor actually asked. Avoid requesting their job title, budget, or contact details before they need to provide them. For example, when someone asks, *“What is the difference between term and permanent life insurance?”*, start with a clear educational comparison. Don't immediately ask them to schedule an appointment. If their question is ambiguous, ask **one relevant clarifying question** and continue from the answer. ## Common types of visitor intent | Intent | What it might sound like | Helpful response | | - | - | - | | **Learning** | “What does this mean?” | Explain the concept in accessible language. | | **Exploring** | “What could work for my situation?” | Discuss options and ask for relevant, non-sensitive context. | | **Comparing** | “How are these approaches different?” | Explain trade-offs without inventing facts. | | **Evaluating fit** | “Can your service help with this?” | Explain scope, limitations, and appropriate next steps. | | **Ready to act** | “Can I book a consultation?” | Point to the available booking or contact option. | | **Seeking current or personal information** | “Which boats are available today?” | Explain when the assistant lacks live data and direct the visitor to a qualified person. | A conversation can move between these intentions. Reassess based on the visitor's latest message rather than assigning a permanent category. ## Adapt without interrogating Good intent-aware conversations follow three principles: 1. **Answer first.** Provide immediate value when the question is clear. 2. **Ask only what helps.** Request context when it will materially improve the next answer. 3. **Respect the visitor's pace.** Offer a next step when it becomes relevant or is explicitly requested. ### Example: leadership consulting **Visitor:** “Our managers are struggling with feedback conversations.” **Assistant:** “A useful starting point is to separate the feedback itself from the way it's delivered. Would it help to explore a simple framework your managers can practice?” **Visitor:** “Yes, and I'd like to see whether your team can train us.” **Assistant:** “I can walk you through the framework and explain the training approach. If you'd like to discuss a program for your team, you can also use the contact or booking option on this page.” The assistant helps first, then introduces human contact after the visitor expresses interest. ## Recognize signals without overinterpreting them Some messages can indicate stronger interest, such as: * Asking how a service works or what is included. * Requesting implementation details, pricing, or a proposal. * Describing a concrete problem or timeframe. * Asking to meet, speak with a specialist, or begin a process. These are **signals to consider**, not proof that someone is ready to buy. A visitor may ask about pricing simply to learn. > [!WARNING] > > Do not infer sensitive personal characteristics, financial capacity, medical status, or purchase readiness from limited conversation data. Avoid asking visitors to share confidential or sensitive information in chat. ## Know when to involve a person A human handoff may be appropriate when the visitor: * Explicitly requests a meeting or personal assistance. * Wants a proposal, quote, or service-specific commitment. * Needs access to live availability, inventory, or information the assistant cannot verify. * Requests individualized guidance requiring professional review. * Reaches a question outside the assistant's defined scope. When a booking or contact button is available on the AI Experience website, direct the visitor to that option. Don't claim that a meeting has been booked unless a connected scheduling process confirms it. ## Configure your assistant for intent-aware conversations Intent-aware behavior is typically defined through your assistant's **system instructions** and supported by accurate knowledge documents. You can adapt this guidance: ```text theme={null} Identify the visitor's immediate goal from their messages. Answer clear questions directly before asking for more information. Ask at most one useful clarifying question at a time when needed. Do not assume that curiosity means purchase intent. Offer relevant next steps naturally, especially when requested. Recommend human assistance when the request requires live data, individualized professional judgment, or action outside your scope. Never invent availability, pricing, bookings, or capabilities. Do not request sensitive personal information. ``` Review this guidance against your business requirements and any applicable professional obligations. ## Test different visitor journeys Use these example prompts to check whether the assistant responds appropriately: | Test prompt | Expected behavior | | - | - | | “I'm just trying to understand the basics.” | Explains without pushing registration or booking unnecessarily. | | “What options should I consider?” | Explores relevant options and asks only necessary questions. | | “How much does your service cost?” | Uses verified pricing information or explains how to obtain it. | | “Can someone call me?” | Directs to the available contact or booking process. | | “What do you have available right now?” | Avoids inventing live inventory or availability. | | “I don't want to share personal details.” | Respects the boundary and continues with general guidance where possible. | If the assistant repeatedly pushes meetings, asks too many questions, or invents information, review both its system instructions and its knowledge sources, then retest. ## Use conversation insights responsibly Visitor questions can help reveal recurring interests, confusing topics, and opportunities to improve your content. Review patterns across conversations without assuming every message represents commercial intent. Use visitor information only in ways consistent with your privacy notices and applicable requirements. ## Related guides * [Design the Conversation](/ai-experience-builder/design-the-conversation) * [System Instructions](/create-and-configure/system-instructions) * [Lead Identification](/conversations-and-leads/lead-identification) * [Conversation Insights](/conversations-and-leads/conversation-insights) * [Human Handoff](/conversations-and-leads/human-handoff) * [Privacy and Access](/conversations-and-leads/privacy-and-access) This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.