As COO of Rozas, Matthew Rozas oversees the systems behind a firm that receives about 1,800 calls a week on its main line. He owns finance, HR, marketing, and operations, so he sees exactly where a new lead can slow down: in the gap between someone asking for help and the intake team getting that request into Clio Grow.
Rozas represents clients nationwide in family-based immigration petitions, asylum cases, removal defense, employment visas, and other immigration matters. The firm also handles misdemeanor and felony cases, as well as complex criminal defense cases throughout Louisiana, along with divorce, child custody and support, and paternity matters.
That made one problem hard to ignore: Rozas was spending too much time typing information from one system into another.
The firm gets about 1,800 calls a week on its main line and between 30 and 70 new leads a day. Before Matthew rebuilt the process, a new lead from the website, Meta, or another marketing channel arrived as an email. Someone on the intake team had to read it, enter the name, phone number, email address, and lead type into Clio Grow, move the lead into the right stage, send a first response, and make the call.
The work was repetitive, and the delays were expensive.
In his Zappy award-winning submission, Matthew estimated that the old process took at least one minute per lead. At 50 leads a day, that meant more than 50 minutes of intake time spent on data entry before anyone could focus on the conversation itself.
A lead becomes a conversation
Matthew and the Rozas team built an automated intake system with Zapier, AI by Zapier, Clio Grow, RingCentral, and Gmail.
Take a new lead from Meta. The lead arrives in the firm's marketing inbox. AI via Zapier reads the message and identifies the case type. A message about bringing a spouse to the United States, for example, can be classified as a family-based immigration matter. The system drafts a response with relevant qualifying questions and a touch of personality, then sends the lead to Clio Grow.
From there, the system sends a text and email through RingCentral and Gmail. It also calls the lead and alerts the intake team that someone new is ready for follow-up. The whole sequence takes about two minutes from form submission.
Matthew described the change simply: the intake team no longer has to wait for a lead, type the lead into Clio Grow, and then make the call. The system creates the record and starts the conversation while the opportunity is still fresh.
Some leads reach the firm via text rather than a form. That used to create another black hole. People text the firm's main number because they are at work, have poor reception, or simply feel more comfortable writing than calling. Those messages were not reaching Clio Grow. The team had to find them, interpret them, and enter the information manually.
The newer workflow brings those messages into the same intake process. Text messages are now working as intended. The team is also building a new version that uses Clio Grow's Marketing Source feature, so the firm's lead sources can become a reportable metric rather than staying buried in a note.
The metric is the conversation
The system is still running as part of the firm's daily intake workflow. Intake specialists use records that are already created and organized. Intake management spends less time on administrative work and more time on quality, conversion, and follow-up.
Matthew estimates that more than 66 hours of administrative time is saved each month, based on a 99.46% successful run rate. Since then, he said, the savings are likely higher as the system has expanded.
The firm has also seen its conversion rate rise from 11.06% in September to 13.58% in May, a 2.52 percentage-point increase. That is roughly a 22.8% improvement. With 50 or more new leads a day, a small percentage change has a real effect.
The intake team's service level moved in the same direction. Matthew said the team was answering just under 60% of calls within the benchmark window in December, January, and February. In 15 or more of the past 20 weeks, more than 80% of calls were answered within 20 seconds.
The system helped because intake specialists were not trying to answer a ringing phone while also processing new leads. They could call people back with the right information already in front of them.
Matthew's goal is straightforward: "We do everything in our power to ensure excellent client communication and rapid response times."
Build what the platform cannot do yet
Lead intake is only the first part of Matthew's roadmap. The lawyers at Rozas already use Claude, ChatGPT, and Clio Vincent, Clio's AI product. But each system has access to different files and capabilities. Claude can work with documents in Google Drive, but not directly with Clio files.
So Zapier keeps the firm's Clio Drive and Google Drive folders synchronized. Lawyers can work with the same documents in the system that fit the task.
Matthew's next idea is an automated case memorandum. When a file opens, he wants Clio Vincent and Claude to independently review the intake forms, case documents, emails, and notes. Each AI would produce a memorandum. The attorney could compare the two outputs, investigate disagreements, and start the case with a clearer view of the issues.
That approach reflects how Matthew thinks about AI. He doesn’t treat one model as an oracle. "I’d like to use seven AIs if I could," he said. "I'd use every single one on the market for the same task each time because all of them are going to be a little wrong and all of them are going to be a little right."
He is also deliberate about where a person stays in the loop. AI can categorize a lead and draft a response. A lawyer should still review a client-facing message before it goes out. That distinction has become part of Matthew's everyday work: looking across the firm for tasks AI can handle, tasks it can assist with, and tasks that still need human judgment.
The Zappy submission began with a faster way to enter a lead. The bigger change is what happened after Matthew saw the pattern. Every repeated task became a question: can the team remove this step, give the work to AI, or make the handoff easier for a person?
At Rozas, a new lead no longer waits in an inbox while someone types. It becomes a conversation.
"The goal is to use AI and Zapier to help us focus on the things that are important."
