How to send a quote in under 60 seconds with AI.

The real test of a small business website isn't how pretty the form is. It's the gap between a customer hitting submit and a priced, personalised quote landing in their inbox. Under 60 seconds is achievable. Under 10 minutes is the baseline. Over an hour is where most businesses sit, and where most of their money is going.

Here's how to build a system that closes that gap. Not a sales pitch — an actual architectural breakdown of what's going on underneath.

Why speed wins

There's a piece of well-cited research (the MIT Lead Response Study, for the curious) that shows the odds of qualifying a lead drop by roughly 80% if you respond after the first five minutes. By 30 minutes, conversion is 21× lower than at one minute. By an hour, you're mostly done.

Every service business I've worked with has felt this but most don't believe the numbers until they run the experiment on their own data. The pattern is universal: fast replies win, slow replies lose, and it doesn't matter how good your quote is if it shows up on day three.

I've written in more detail about the real cost of slow lead response if you want the full picture.

The architecture, in five pieces

A real-time quote system has five components. They can live in different tools or one — what matters is that the handoffs between them are instant.

1. A structured intake form

Not a free-text "describe your project" box. A proper form with the specific fields your pricing depends on — job type, square metreage, materials, urgency, site access, postcode. The form asks only what it needs. No "phone number" unless it's genuinely required.

This is the step most businesses skip. They've got a generic contact form that asks for a name and a message, and then they wonder why quoting takes forever. The AI can only move fast on information that's actually captured.

2. A webhook firing on submission

The moment the form is submitted, a webhook fires off to your quoting agent. This is a 10-line piece of plumbing. It matters because it removes the delay between "customer submitted" and "system starts working." Polling an inbox every 15 minutes is a bad architecture for a speed game.

3. An AI agent with your pricing rules

This is the brain. It takes the structured input, runs it through your pricing logic (this is where it gets interesting — you can encode surprisingly complex rules), looks up any relevant context (customer history, postcode-based pricing, seasonal adjustments), and drafts a quote in your business's voice.

The common failure mode here is giving the AI too much freedom. Don't ask an LLM to "decide what to charge" — encode your pricing rules explicitly and let the AI handle the language around them. The AI writes. Your business rules decide.

4. A templated email generator

Same reason your invoices look the same — consistency builds trust. A quote email has a specific structure: scope, inclusions, exclusions, price, timeline, next steps. The AI fills in the variables and renders the email. Nothing flashy.

5. A CC to you, and a log to your CRM

The moment the email goes to the customer, a copy goes to you, with a one-line summary and the ability to stop it if something looked wrong. The full record lands in your CRM so it's ready when the customer replies.

What most first attempts get wrong

I've seen a lot of small businesses try to build this themselves with Zapier and ChatGPT and end up frustrated. The common failures:

1. Asking the LLM to do pricing. LLMs don't do arithmetic reliably and aren't supposed to. Your pricing rules need to live in code or a spreadsheet lookup, with the LLM only responsible for language around the numbers.
2. No review step for edge cases. A system that sends every quote without any human oversight will eventually send a weird one. Build a "hold for review" branch for anything outside normal parameters.
3. Over-engineering. A beautiful PDF quote with drone footage is worse than a plain email that arrives in 30 seconds. Speed beats polish, always.
4. Skipping the intake form redesign. The form is the input. Bad input, bad output. The single highest-leverage thing you can do here is rework the form.
5. Not integrating with your calendar. A quote that offers "let me know when suits" is a worse quote than one that says "I've held Thursday 10am — reply YES to confirm." The AI can check availability in real time.

A realistic timeline for getting this live

This is the kind of system I build during the free 14-day period. Rough sequence:

• Days 1–3: Understand your current pricing, edge cases, and what a good quote from your business looks like. Rewrite the intake form.
• Days 4–8: Build the pricing logic, wire the webhook, draft email templates, and do a first pass of the AI prompt.
• Days 9–11: Test with real past quotes to make sure the system produces outputs you'd have sent yourself.
• Days 12–14: Go live with review mode on. You approve the first week of quotes. Once you trust it, auto-send kicks in.

Full breakdown in the 14-day build post.

What "under 60 seconds" actually looks like

A customer submits the form at 11:47pm from their couch. The form fires a webhook at 11:47:02. The AI runs, drafts a quote matching your pricing logic by 11:47:18. The email renders and sends by 11:47:24. They receive it at 11:47:32. They're still on your site.

From their side: fastest tradie they've ever contacted, clearly organised, probably trustworthy. From your side: you're asleep. You wake up, there's a quote out and a reply waiting.

This is the standard now. If you're not at this speed, somebody else's AI is beating you to every lead.

Want this for your business?

The short version of the offer: email me one sentence about your current quoting process. If it's a fit, I'll build this for you during a 14-day free build. You'll see the first live quote go out around day 10, and once you trust the outputs, the whole thing runs on autopilot.

More on why I build the first one free, and on who I am.

— Steve the AI Guy

---