The founding attorney was answering WhatsApp at 11pm. $1M in ad spend. Zero conversions tracked.
Grupo Lyown, a Miami-based law firm with operations in Colombia, was losing lawyer billable hours to manual WhatsApp intake. Work-Smart built Victoria, an AI agent that pre-qualifies leads on WhatsApp, books meetings, and tracks everything in CRM. Lawyers only see leads ready to talk.
Lawyers doing intake work. No CRM. No idea what was happening.
Grupo Lyown is a legal consulting firm that helps foreign entrepreneurs navigate the complexities of doing business in the United States, trademark registration, company formation (LLC/Corp), immigration visas (E-2), tax compliance, and cross-border legal strategy. Their clients are predominantly from Colombia and Latin America, many of them first-time business owners in the US.
The firm had clients. What they didn't have was a system to find, qualify, and convert new ones without burning lawyer time on it.
Their lead flow worked like this: someone would find them through a Google search or a Meta ad, land on a website that barely loaded, performance score: 8 out of 100. That's not a typo. If they persisted, they'd submit a general inquiry. That inquiry went to a shared email inbox. A lawyer would then respond manually on WhatsApp, sometimes within hours, sometimes the next day. The conversation would bounce between questions about services, pricing, and scheduling. Eventually, maybe, a meeting would get booked.
The lawyers themselves were spending time on WhatsApp doing what was essentially intake work, answering the same qualifying questions for every lead. What city are you in? What's your budget? What service do you need? Is your company already formed? These are not questions that require a law degree to ask. But the lawyers were asking them because nobody else was.
Meanwhile, there was no CRM. Lead information lived in WhatsApp conversations and memory. There was no way to know how many leads came in last month, where they came from, which campaigns drove them, or what happened to them after the first message. When Javier wanted to understand his sales pipeline, there was nothing to look at.
The problem wasn't traffic. The breakdown was mid-funnel.
Lyown's problem wasn't traffic. Meta and Google ads were generating leads. WhatsApp conversations were being created. The AI agent (once built) was responding correctly. The breakdown was mid-funnel: leads started conversations but didn't schedule meetings. Three structural gaps drove the failure.
Data Layer
No CRM. No lead tracking. No way to know where leads came from, what they asked, or what happened to them. Customer data lived in WhatsApp threads and scattered tools. When Javier wanted to understand his pipeline, there was nothing to look at.
Automation Layer
The lead capture flow was manual at every step. Each inquiry required a human to respond, qualify, and schedule. For a legal firm where every hour is billable time, this was a direct business cost, not just an inconvenience. The AI agent (once built) asked too many questions before sending the booking link. When a lawyer jumped in manually via WhatsApp, the AI would continue talking, creating overlapping and confusing messages.
Visibility Layer
Javier had no way to see what was happening in his digital operations. How many leads this month? Which campaign is converting? What percentage of conversations result in booked meetings? None of these questions had answers. The analytics dashboard showed 8 total conversations, 0% appointment rate, but there was no system to diagnose why.
There was no conversation governance defining when the AI speaks, when the human intervenes, and when the AI should pause. That single structural gap, human and AI talking over each other, was breaking a significant portion of the conversation flow.
Meeting booking rate = the percentage of WhatsApp conversations Victoria started that ended with a booked meeting on a lawyer's calendar.
"Before Victoria, our lawyers were answering the same WhatsApp questions every day. Now they only see leads that are ready to talk. We finally have a sales pipeline we can measure."
Javier Rodriguez Gamarra, Founder, Grupo Lyown
Website Redesign + Landing Pages
The website was rebuilt from scratch, designed to turn visitors into leads, SEO-ready, with technical setup so AI tools like ChatGPT can find and recommend Lyown. Performance went from a score of 8 to a functional, fast-loading site. Separate landing pages were created for specific ad campaigns, intellectual property consultations, trademark registration for foreign businesses, E-2 visa support. Each landing page speaks directly to one service and one ICP, with clear calls to action that feed into the automation flow.
Victoria. WhatsApp AI Agent
Victoria is an AI agent that responds to incoming WhatsApp inquiries instantly. When someone submits a form on the website or clicks through from an ad, Victoria sends the first message within seconds. Victoria does four things: responds immediately (no waiting for a human), personalizes the conversation based on which service the lead is interested in, asks only the necessary qualifying questions (city, budget range, company status), and sends the Calendly booking link early in the conversation, not after 10 questions. When a lawyer needs to intervene manually, Victoria pauses automatically. The AI resumes only from the admin panel or by command. No more overlapping messages, no broken conversation flow. Victoria also sends automated reminders. 24 hours and 48 hours before the meeting. If the lead doesn't book, Victoria follows up twice, then stops. Every interaction is logged.
CRM and Lead Dashboard
A custom admin panel replaced the scattered tools. Every lead is tracked: name, contact info, source (which ad campaign, which landing page, organic or paid), service requested, conversation history, and status (new, active, meeting booked, disqualified). The panel shows all leads in one place, their source and attribution, the service they asked about, files shared via WhatsApp, the full interaction history, and current status. This replaces Excel and manual tracking.
Analytics and AI Visibility Foundation
Google Analytics, Google Search Console, and Bing Webmaster were set up and configured. Sitemap submitted to both Google and Bing (with Claude/ChatGPT indexing through Bing). After the domain change, we ensured search engines could find and list all of Lyown's pages. We also identified organic authority opportunities across Reddit and legal forums, channels where potential clients research before hiring.
Monthly Retained Operations
Two tracks: WhatsApp AI agent monitoring (conversation optimization, campaign-specific adjustments, stability, lead supervision with manual intervention when needed) and website + AI visibility (Analytics and Search Console review, landing page optimization based on keyword research, new content creation, Reddit/Quora authority posting, and marketing team collaboration).
Meeting booking rate
0%
→42%
Lead response
Hours/days
→Seconds
Build time
,
→60 days
Attorney nights
11pm WhatsApp
→Automated
- ▸The system is designed to compound. Every month, the AI agent gets smarter (conversation flows optimized based on real data), the landing pages get more targeted (keyword research drives new campaign pages), and the authority content accumulates.
- ▸The conversion diagnosis revealed that the bottleneck was never traffic, it was the mid-funnel orchestration. Fixing the booking flow (sending the Calendly link earlier), adding conversation governance (AI pauses when human speaks), and shortening the qualification sequence increased meeting bookings without increasing ad spend.
- ▸Lead information that used to live in WhatsApp conversations and memory is now in a structured CRM with full source attribution, service history, and conversation logs.
- ▸Website performance went from a score of 8/100 (12.4-second load time) to a rebuilt, conversion-optimized, fast-loading site that search engines can index.
- ▸Monthly retained operations mean the system keeps improving, not just maintaining. Each month adds tuning, content, and optimization based on real data.
Questions About This Case Study
The initial build, website, Victoria AI agent, CRM, analytics setup, was completed in approximately 8 weeks. The AI agent was in production within the first month. The attorneys stopped checking WhatsApp for leads within the first week. Victoria handled it. That's the adoption test: when the team trusts the system enough to stop doing the work themselves. The system continues to evolve monthly through the retained operations engagement.
This was a complete digital growth system build: website redesign, Victoria AI agent, custom CRM, analytics setup, and landing pages. The engagement was structured in phases with clear deliverables for each component. The ongoing retainer covers AI agent monitoring, conversation optimization, website maintenance, landing page updates, and AI visibility. Pricing depends on practice areas and complexity, the AI Ops Audit scopes exactly what you need.
Yes. The pattern is consistent across legal firms in this size range: lawyers spending time on intake, no CRM, no lead tracking, no system for qualifying before the first meeting. The specific qualifying questions and conversation flows change based on your practice areas, but the infrastructure is the same. The key insight from working with legal firms: lawyers are understandably skeptical of AI for legal reasoning (their license is at risk). This system doesn't touch legal work. It handles the admin and intake workflow that eats their billable hours.
No. Victoria connects to your existing WhatsApp business account, Calendly, and whatever CRM or tools you already use. If you don't have a CRM, we build one. The system sits on top of your current workflow and automates the parts that don't require legal expertise.
Monthly retained operations. The AI agent is monitored and optimized, conversations get smoother, qualifying questions get refined based on which leads actually convert. Landing pages are updated based on keyword and campaign performance. New content is created to build organic authority. You get a monthly report showing leads, sources, conversions, and recommendations.
Victoria filters by asking qualifying questions before engagement. City, budget, service need, company status. Messages that don't match criteria get flagged but don't consume attorney time. The system also detects spam patterns, multiple messages from a single user, WhatsApp submissions that bypass the form.
The firm's ad spend stayed the same. Before Victoria, conversion was effectively 0%, leads came in but nobody followed up fast enough. After Victoria, 11 out of 57 leads converted to confirmed meetings within the first month. Response time dropped to 47 seconds. The ROI wasn't from more leads, it was from actually converting the leads they were already paying for.
If your firm's lawyers are spending time on WhatsApp answering the same qualifying questions instead of practicing law, you're paying for human intake work that a system should handle.
The AI Ops Audit is how every engagement starts: 2-3 weeks, and you'll know exactly where your intake process is leaking and what it takes to fix it.