Overview

If you already use Claude or ChatGPT every day, you can point it at your LeadBind account and just ask.

"What should I work on today?"

"Book Maria for Thursday afternoon."

"Which leads went cold this week?"

It reads your real leads, your real calls, and your real calendar. It is not a chatbot trained on our help articles. It is your account.

Why this matters more than it sounds

Every other voice AI product is a closed dashboard. You log into theirs, you look at their charts, you leave.

LeadBind is different in one specific way: your assistant already has your other tools connected. Your calendar. Your email. Your invoicing. When LeadBind joins that set, your AI can do things neither of us built:

• "Book this lead, then email me their file before the appointment."
• "Anyone I called twice with no answer, add them to my Monday list."
• "Look at last month's booked calls and tell me which ad source is actually worth the money."

We did not build an email integration for that first one. Your assistant did it, because it holds both tools at once. That is the whole idea.

What you can build

A few things that take one sentence to your assistant, or one line in a terminal. All of them run against your real account.

Drive the phone from your chat window. Your assistant can dial a lead, let the AI agent book them, and the call lands back in your history.

"Call the lead I just added and book them for Tuesday afternoon."

leadbind call --phone=8325550134 --name="Maria Lopez"

Text a deposit link without leaving the terminal. The money settles into your own connected Stripe.

"Send Maria a $250 deposit link for the roof job."

leadbind bill --amount=250 --for="Roof deposit" --lead=ld_abc123

Turn a cold list warm every morning. Ask for work, not records, and let the assistant act on it.

"Anyone I called twice with no answer this week, put them on a follow-up sequence and call the freshest three back now."

Build a whole account by talking to it. The config tools let an assistant stand up calendars, availability, pipeline stages, custom fields, and drip sequences. Have it show you the plan first:

"Set up a roofing pipeline: an inspection calendar with 30-minute slots, a three-touch no-answer follow-up, and a 'roof type' field. Show me the dry run before you write anything."

Cross the line into your other tools. This is the part neither of us built. If your assistant also holds your calendar or email:

"Book my 2pm from LeadBind, then draft the confirmation email with directions to their address."

Every one of these maps to tools on the reference page. Placing calls needs the dial permission; the rest need read or write.

Getting started

1. Connect an assistant: create a key, paste one config block. About two minutes.
2. Tools reference: every question it can answer and every action it can take.
3. Keys and permissions: what each permission allows, and how to revoke.
4. What we send and what we never send: read this one before you connect.

Prefer a terminal? See the command line.

The short version of the safety story

Full detail is on the data page, but the three things worth knowing now:

• Call recordings and transcripts never go to your assistant. Not through the API, not the CLI, not MCP. Recorded conversations carry consent rules that vary by state, and your customer never agreed to an AI vendor holding them. They stay in LeadBind.
• Placing calls is a separate permission and it is off by default. Reading your leads and dialing real people are not the same risk, so they are not the same switch.
• Your leads' names and numbers do go to whichever assistant you pick, because that is how it answers you. You are choosing that provider, and that is a real decision. Make it on purpose.

Docs your assistant can read

This documentation is built for machines as well as people.

• Every page has a raw markdown twin: add .md to any URL.
• docs.leadbind.org/llms.txt lists every page.

So an assistant connected to LeadBind can answer its own setup questions rather than sending you back here.