AI meeting prep doc: a one-page brief in your Drive before every booking

What lands in your Drive, exactly

A new Google Doc titled Prep — {Lead Name} — {Date} is created in a folder called “AutoSync Meeting Prep” on your Drive. The doc has six sections, all auto-filled:

• The booking facts. Lead name, email, when they booked it, when the meeting is, what they wrote in the booking notes.
• What we infer about them. Email-domain → likely company. Free-text booking notes parsed for what they're actually asking about.
• Three questions to ask. AI-generated, specific to what they wrote in the notes — not generic “tell me about your business” filler.
• Two likely objections. The most common pushback in your industry, drawn from your AutoSync intake answers about what you sell.
• One opener. A concrete first sentence that references their actual situation, not “Hi {Name}, thanks for jumping on a call.”
• Your own action items. A blank checklist you fill in during the call. The doc stays in your Drive forever as a per-meeting log.

Why “just paste their email into ChatGPT” fails

Three frictions that kill this in practice:

• You forget. The booking comes in at 11pm. You see it at 8am. The meeting's at 9am. You don't paste anything; you just dial in.
• Context every time. ChatGPT doesn't know what you sell, who you sell to, or what your typical objections are. You either type that every time or get generic output.
• It lives in a tab. Five days later when you want to remember what you discussed, you can't find the chat. The Google Doc sticks around in a folder you actually look in.

The model under the hood (and what it costs)

We use Claude Haiku 4.5 — Anthropic's fastest, smallest model. Each prep doc costs us roughly $0.001 in API spend (~$1 per 1,000 bookings). We don't bill you per doc; it's bundled into the Quick Wins tier and above.

Why Haiku and not GPT-4 / Claude Opus: prep docs don't need genius-tier reasoning, they need fast, structured output. Haiku gives us a 2-second draft for the price of a fraction of a penny. Cheap enough that a customer who books 200 calls a month costs us 20 cents in AI.

What changes about how you run the call

Concrete behavioral shift — the 8 minutes between “sit down” and “dial in” goes from cold to warm. You read a one-page summary, jot two notes, and walk in already aimed at the conversation that matters.

Side benefit: the docs accumulate. After 50 bookings you have 50 prep docs sitting in one folder. Search any name, see what was on your mind going in + what you wrote during the call. Better than any CRM that you'd have to manually populate.

What it doesn't do

Does not join the call. No transcription, no “AI notetaker” bot. You type your own action items during the meeting.

Does not enrich from external data. No LinkedIn scrape, no Crunchbase lookup. Inference is from email domain + their booking notes only — privacy-preserving by default.

Does not work for meetings booked outside AutoSync. If someone emails you and you scheduled it manually, no prep doc. We only fire on the AutoSync booking flow.

Does not edit existing docs. One doc per booking, by design. If a lead reschedules, you get a second prep doc; the first stays as a record.