Questions, answered as far as they can be

The target is a first message inside about 30 seconds of the lead arriving. Real-world timing depends on how quickly your lead source hands the inquiry over, which is often the slower half. The speed-to-lead test walks you through measuring your own end-to-end time.

Web chat and email work today. SMS reply is coming soon, and WhatsApp would go through the official business API only. Voice is coming later, after the messaging channels are settled.

No. The first message says it is an assistant working on your behalf and that a person will follow up. That is a default, not an option.

It says it does not know and offers to get you. It answers from the material you give it: your listing facts, your service area, your standard answers. It is built to decline rather than guess at a price, a school rating or a condition.

Every thread is readable in full, and every conversation writes a transcript note back to your CRM. When a reply was wrong, you edit the script or the source material and the change applies to the next lead.

No. An opt-out ends every sequence for that contact immediately and is recorded. A human reply also ends the sequence and hands the thread to you.

We quote per account, because the right plan depends on how many agents you have, how many lead sources you want connected, and how much volume comes through them. Tell us those three things and you get a number back.

Not right now. Accounts are not open yet. You can join the waitlist, and the live demo and all seven free tools work without an account.

One person in one thread over a rolling window, no matter how many messages go back and forth. A lead who replies four times is one conversation, not four.

Web chat and email are included in every plan. SMS reply is coming soon, so there are no per-message costs to pass through yet.

Monthly plans cancel at the end of the current period and keep working until then. Annual terms are set in your quote.

You get a notice before the limit, not after. Replies keep going out while you decide whether to raise the limit or let the sequence pause.

Follow Up Boss, kvCORE and BoldTrail, and Zillow lead feeds. Each integration page lists what it reads, what it writes, and what it cannot do.

Four objects on every conversation: the contact, a note holding the transcript, a task with the next step, and the appointment if one was booked. You can turn any of the four off.

Matching runs on phone number and email before anything is created. An existing contact gets a new note instead of a second record. Matches are listed in the sync log so you can check them.

Consent is yours to collect and yours to prove. The product records opt-ins and opt-outs, applies quiet hours by contact time zone, and includes opt-out language on messaging channels. That is general information rather than legal advice. Ask your broker or your counsel about your own setup.

Qualifying never scores or routes anyone by a protected characteristic, and listing drafts come with a review checklist for phrasing that describes people rather than property. Drafts are drafts. A licensed human has to read every one before it goes out.

Voice is not part of v1, so nothing is recorded today. If a voice phase ships, recording consent rules vary by state and the feature would be off until you turn it on per market.

You do. Export is available, and deletion removes the thread and its transcripts. Retention defaults and processor list: see the privacy page.

No. Your conversations are not used to train or fine-tune a model. Where a third-party model provider drafts a reply, the request is sent with training opt-out where the provider offers one. The privacy page describes the categories of processor involved.