Lead Capture and Human Handoff in the AI Receptionist

The AI receptionist’s job has two halves. The first half is answering — knowing the business well enough to handle most questions. The second half is capturing or escalating — recognizing the moment a conversation needs to result in a lead or a human handoff, and executing that transition smoothly. This article covers the second half.

What “lead capture” means in chat

The receptionist captures a lead when it collects the visitor’s contact information and creates a pipeline entry — the same kind of entry a form submission would create. From the agency’s perspective, a chat-captured lead is indistinguishable from a form-captured one. Same card, same pipeline, same notification email.

What changes is the experience on the visitor’s side. Forms are a one-shot transaction; chat is a conversation. The receptionist can ask for information conversationally, in response to what the visitor said, in the moment when they’re most engaged. That context usually produces higher-quality leads than a generic contact form.

The capture flow, in practice

Lead capture happens when the receptionist decides the conversation has reached a point where contact info is warranted. The trigger is contextual, not scripted:

• A visitor asks a question that needs a real person’s answer (“Can someone call me back to discuss my case?”)
• A visitor expresses clear intent (“I’d like to book a consultation”)
• A visitor describes a situation the receptionist knows is qualifying (“I need a will drafted”)
• The visitor has spent several minutes in conversation and shown sustained interest
• A visitor asks for something the receptionist can’t book directly (custom service, complex coordination)

When any of these surface, the receptionist asks for contact info naturally:

“Got it — I’ll have an attorney reach out about your matter. Can I get your name, email, and the best phone number to reach you?”

If the visitor provides the information, the receptionist immediately:

1. Stores the visitor’s name, email, and phone in the platform’s identified-visitor record
2. Creates a lead pipeline card with the conversation linked
3. Fires the notification email to the configured email address (the one you set during receptionist configuration)
4. Sends the visitor a small confirmation (“Thanks Sarah — Anna from the team will be in touch within the hour”)

The whole capture happens in a few message exchanges. No popups, no form interruption.

When capture happens, when it doesn’t

Capture isn’t automatic. The receptionist is configured to ask for contact info when it makes sense for the conversation, not as soon as possible. Asking too early kills the conversation; asking too late means missed leads.

The behavior is tunable through the persona field and house rules in your knowledge base (see Train Your Receptionist on Your Website).

To capture more aggressively, add to house rules:

After the second question from a visitor, gently ask if they'd
like a follow-up call. Don't push if they decline, but always
offer at least once per conversation.

To capture only on high-intent signals, add:

Do not ask for contact info unless the visitor has explicitly
expressed intent (booking, callback request, case description).
Casual browsers and pricing-only inquiries should not be asked
for contact details.

What the visitor experiences during capture

A small but important detail: the receptionist captures contact info before taking the next action, not after. If the visitor says “book me a consultation,” the receptionist asks for contact info, then offers slots. This is intentional — capturing first means the lead exists even if the visitor abandons before booking.

Three sub-flows emerge from this:

• Capture + book. Visitor asks for an appointment, receptionist captures details, then offers slots. Pipeline gets a lead + booking record.
• Capture + escalate. Visitor asks something the receptionist can’t handle, receptionist captures details and confirms a callback. Pipeline gets a lead + escalation note.
• Capture only. Visitor asks general questions, receptionist answers and at some point offers a follow-up, captures details. Pipeline gets a lead without a specific next-step.

All three count as captures. The agency sees them all the same way in the pipeline.

Human handoff — when AI passes the conversation to a human

Capture creates a lead the agency follows up on later. Handoff transfers the conversation to a human right now.

There are two ways a handoff happens.

Visitor explicitly requests it

The visitor says “I want to talk to a real person” or “Can a human help me?” or “Are you a bot?” — anything that signals they want a human. The receptionist recognizes this and triggers the handoff tool. The conversation transitions immediately.

The visitor sees a message like:

“Of course — let me get someone for you. Can I quickly get your name and email so they can pick up the conversation?”

The visitor provides contact info, the handoff completes, and the visitor sees:

“Thanks Sarah! Anna from the team will be with you shortly. If we don’t reach you in the next 5 minutes, we’ll send a follow-up email to [email protected].”

Receptionist hits a wall

The visitor asks something the receptionist genuinely can’t answer — something outside the knowledge base, something policy-bound, something requiring judgment. The receptionist initiates the handoff itself.

The visitor sees:

“That’s something I’d want our team to answer properly. Can I get your name and email so [Attorney Name] can follow up directly?”

This is the safer pattern — the receptionist gracefully exits situations where it might make things up.

What happens on the agency side

The moment a handoff is initiated, an email lands in the configured escalation inbox containing:

• The full conversation transcript, formatted for readability
• A summary the AI generated about what the visitor wants
• The visitor’s contact info
• A direct link to the conversation in the dashboard

The conversation in the dashboard is now flagged as escalated. It appears in a separate “Needs response” section on the Receptionist page until a human responds.

Taking over a conversation

The agency can take over an escalated conversation directly in the dashboard. From the Receptionist page, click any escalated conversation. The conversation panel shows the full history. At the top, a “Take over” button.

Clicking “Take over” puts the conversation in human mode. From this point forward:

• Any message the visitor sends is routed to your email instead of the AI
• You reply by sending email back through the same thread — your reply lands in the chat as a message
• The visitor sees your reply as coming from a real person, with your name (not the AI’s persona)
• The AI does not generate any more responses unless you click “Return to AI”

Email-as-chat is intentional. It means humans can respond from wherever they handle email — phone, desktop, inbox app — without needing to log into Captivar. The conversation continues even if the agency is mid-meeting; they just need to check their inbox.

Configuring escalation

Three settings shape escalation behavior, all on the Receptionist configuration page:

Escalation email

The address that receives handoff notifications. This is the inbox someone watches. Set it to the person who responds fastest — owner, lead intake staff, the on-call rotation address. Default is the site’s notify_email.

Auto-escalate on unanswered questions

Off by default. When on, the receptionist initiates a handoff any time it can’t find an answer in the knowledge base — even if the visitor didn’t ask for a human.

Turn this on if you want every gap in the knowledge base to surface to a real person. Turn it off if you’d rather have the receptionist say “I don’t know” gracefully without escalating to a human every time.

Most agencies leave this off and rely on visitor-initiated handoffs.

Response time expectation

A free-text setting like “within 1 hour” or “by end of business day.” The receptionist uses this verbatim in the message it sends the visitor at handoff. Set realistically — under-promising and over-delivering builds trust; over-promising and under-delivering breaks it.

For agencies actively staffed during business hours, “within 1 hour” is a reasonable expectation. For after-hours handoffs, the message automatically adapts to “by the next business day” without you needing to configure it.

Off-hours handoffs

If a visitor requests a handoff outside business hours, the receptionist captures their info and confirms a next-business-day response. The escalation email still fires immediately — what changes is the promise to the visitor.

This is configured implicitly by your business hours (set during receptionist configuration). The receptionist knows the time, knows the schedule, and adapts the language accordingly. You don’t need to set anything else.

The lead pipeline view

Every captured lead and every escalated conversation appears on the lead pipeline (covered in Working the Lead Pipeline). The card shows:

• The visitor’s name and contact info
• The source channel — “Chat — receptionist” or “Chat — escalated”
• A small icon indicating whether it’s a regular capture or an escalation
• The pipeline status (defaults to new)

For escalated conversations, the card includes a small badge: “Awaiting human reply.” This stays until you reply from the dashboard or via email reply.

Three patterns that work

Operational patterns from agencies who use the system well:

Set a 1-hour response SLA and stick to it. The receptionist promises “within 1 hour”; the team delivers. This single discipline accounts for most of the difference between sites that convert and sites that don’t.

Use auto-escalate sparingly. Most teams leave it off because it floods the escalation inbox. Better to expand the knowledge base than escalate every gap.

Read escalated conversations even after a human responds. They’re free training data for what the receptionist should know but didn’t. Every escalation is a knowledge base improvement opportunity.

What happens next

You now have the complete operational picture: the receptionist captures, the pipeline organizes, the team responds. The remaining doc that ties everything together is roles, plans, and access — the conceptual layer underneath. That’s next.