How I Sold My First $1,800 AI Automation to a Law Firm Full Deal Breakdown

I want to walk you through a complete client win I closed for our AI automation agency. This post covers the sales playbook I used, the discovery questions that unlocked the opportunity, and the technical system we built to automate a cold email lead generation process for a boutique mediation law firm in Austin, Texas. If you want to clone this workflow and sell it to other small law firms, you will find step by step guidance, the logic behind each decision, and the exact pieces that make the automation work.

Join our free AI Automation Community to download the n8n template, sales process, and other materials we used to close this deal.

What I Covered

• How we sourced and qualified the lead
• The discovery approach that captures real workflows
• The four step sales process we use
• Technical design for a lead generation automation that scrapes directories and firm sites
• How we use Firecrawl, n8n, AI evaluation, and Google Sheets and Docs to create ready to send outreach

Sales Process Overview

To win customers consistently you need a repeatable sales playbook. Our process has four clear steps:

1. Lead sourcing
2. Qualification
3. Discovery
4. Close

1. Lead sourcing

Start with warm leads when your agency is early stage. Look inside your network for professionals who run manual workflows. For us, a friend who runs a mediation practice in Austin fit perfectly. Warm leads shorten the sales cycle and give you an initial case study that helps you land the next client.

2. Qualification

Run a short 15 to 30 minute qualification call with this goal: confirm a real problem and that it's worth going deeper. Ask four core questions on every qualification call:

• Is there clear fit between your offering and their work?
• Is the problem urgent or time bound?
• Are you speaking to a decision maker?
• Do you have access to the data needed to automate the work?

Keep the call focused. If the answers align, schedule a full discovery session. If not, close the conversation and move on.

3. Discovery

Discovery is where you capture the actual workflow as it runs today. The rule is simple: ask to see the real work, not a summarized version. Have the prospect share their screen and walk you step by step. Your job is to collect concrete details that the automation must reproduce.

For the mediation law firm we asked questions such as which filters they use in the Texas lawyers directory, how they verify emails, what makes a target attorney a good fit, where they store notes, and how they track replies. These small details become the backbone of the automation logic.

4. Close

Present a clear offer after discovery. We typically price this type of build around $1,800 for the initial build and $400 per month for maintenance. For the first client we accepted a reduced rate in exchange for a testimonial. Document the proposal and scope in writing and get a signed agreement before starting the build.

Automation Design Summary

We built a two part automation system to replicate the lawyer research and outreach process. The design goals were:

• Source relevant law firms from a public directory
• Extract attorney profile pages from each firm site
• Evaluate profile relevance using an LLM
• Extract contact fields and areas of practice using Firecrawl
• Save validated leads into Google Sheets and draft outreach emails as Google Docs

Component 1 Directory Scraper

This component accepts a directory page URL and returns a structured list of law firms and cities. Key elements:

• Use Firecrawl to scrape the directory page and extract the firm name and city into JSON
• Split the result into individual firm items for processing
• Check Google Sheets to avoid duplicates
• Search the web to find the firm website and verify it is an individual firm site
• Append the firm to the law firms sheet and pass it to the next component when valid

Firecrawl's extract feature lets you define a custom schema so you get a clean array of objects rather than raw text. This reduces parsing work later.

Component 2 Firm Profile Crawler and Lead Extractor

This is the heart of the system. It processes a single firm website and yields a set of validated leads. The subworkflow runs once per valid firm detected by the directory scraper. Core steps:

1. Normalize the incoming input and extract the domain
2. Use Firecrawl search restricted with site colon to list candidate profile pages
3. Filter search results with an LLM to keep only individual attorney pages
4. Scrape each profile page with Firecrawl extract schema for name, title, email, phone, and areas of practice
5. Check Google Sheets to avoid duplicate person entries
6. Run an AI evaluation to determine if the profile matches the target persona
7. Generate a ready to send email draft as a Google Doc and link it in the sheet
8. Append the validated lead row to the people sheet

Use a broad but targeted keyword set when searching for profile pages. For lawyer sites we included keywords such as attorney, partner, associate, and profile. Then rely on an LLM to remove false positives such as practice area pages or news posts.

LLM Evaluation Checks

There are two critical AI checks along the pipeline:

• Is this a law firm website based on title, URL, and meta description
• Is this an individual attorney profile based on search result data

Later we run a profile evaluation where the LLM sees the scraped profile content, the listed practice areas, and the candidate's role. The LLM returns a boolean indicating whether the attorney is a match for the mediator outreach criteria. This is how we avoid outreach to irrelevant profiles.

Structured Profile Extraction with Firecrawl

For each confirmed profile page we ask Firecrawl to return both markdown text and a JSON object with the fields we need. The JSON schema includes:

• First name
• Last name
• Email
• Phone
• Position
• Firm
• Areas of practice array

This approach prevents brittle HTML parsing logic. If the site layout changes, Firecrawl's extract prompt can be tuned quickly without rewriting code nodes.

Google Docs Email Drafting

Each validated lead gets an email draft generated using a template with placeholders. We convert the template from markdown to HTML, create a Google Doc via API, and store the doc ID in the sheet next to the lead row. That gives the client a copy paste email for outreach.

Operational Notes and Best Practices

• Start warm when building your first clients. Use your network for initial leads.
• Capture reality in discovery by asking to see real workflows and files.
• Make deduplication explicit in your automation to save API quota and prevent repeated outreach.
• Use search then scrape rather than broad scraping to limit noise from external directories.
• Keep the AI prompts explicit so you get deterministic boolean outputs your workflow can act on.
• Log everything in Google Sheets so the client can audit leads and downloads.

How to Sell This to Similar Professional Firms

Package this as a repeatable product for local professionals who rely on referrals and targeted outreach. The sales message should be simple and direct:

• We automate the research and outreach process that currently takes hours of manual time
• We deliver a steady stream of validated leads with ready to send email drafts
• We provide a low friction trial for the first client in exchange for feedback and a testimonial

In qualification calls lead with a quick, four question checklist on fit, urgency, data access, and decision maker. If all four are yes, move to discovery. If any are no, end the process.

What You Get When You Join AI Automation Mastery

Members gain access to the full n8n JSON template for this lead generation workflow plus step by step notes for adapting it to other niches. Use the template to speed up your first build and to show a working demo to prospects during discovery.

Conclusion

This project shows how a small, repeatable sales process combined with a targeted two part automation can convert a manual cold email workflow into a predictable lead pipeline. Start with warm leads, run short qualification calls, capture the real workflow in discovery, and then build a clean automation that integrates Firecrawl, an LLM, Google Sheets, and Google Docs.

Join AI Automation Mastery to download the n8n template and get help deploying this workflow for your clients.