Improving Lead Entry: Scaling by 60% with AI

Everyone is talking about AI right now, and I'm fascinated by how many conversations start with the technology instead of the problem.

I think that's backwards.

Our lead entry team used to work out of a shared inbox. Insurance brokers would send quote requests to multiple carriers, usually by email. Every request contained the same basic information, but every broker organized it differently. Some information was buried in the body of the email. Some was attached in spreadsheets. Some lived in PDFs.

The job was to gather all of that information, interpret it, and enter it into our systems so the quoting process could begin.

It was painful.

Ownership wasn't always clear, requests could get buried, and we had very little visibility into what was actually happening.

Before I touched AI, I moved the team into Salesforce Cases. Suddenly, we could see workloads, assign ownership, and understand where requests were getting stuck.

Process first.

Then came Agentforce.

There was a lot of trial and error. Some ideas worked. Some absolutely didn't. Eventually, we landed on an AI and human workflow that improved the team's experience without removing the people responsible for the work.

Then I found another opportunity.

About a quarter of our requests followed a standardized format. Those were simple enough to handle with a Python script.

The remaining requests were much more complicated. The information was still there, but traditional automation couldn't reliably identify it because the formatting varied so much.

That's where an OpenAI API connection came in.

And no, we didn't go out and buy another platform. We simply paid to access the API and built exactly what we needed.

More importantly, we didn't try to replace people.

That would have been a mistake.

Instead, we built checkpoints throughout the process where employees could review the output, validate the information, and step in when something didn't look right.

AI wasn't making decisions in isolation. People and AI were working together.

Over time, we increased our weekly processing capacity by 60%.

That's a huge improvement, but it still wasn't enough. Our long-term goal requires us to more than double our current throughput. That's why contractors are still part of the strategy.

AI isn't a miracle drug. Sometimes the answer is a new tool. Sometimes it's a process redesign. Sometimes it's automation. Sometimes it's AI. And sometimes it's recognizing that you're solving a capacity problem, not a technology problem.

The solution wasn't Salesforce. The solution wasn't Python. The solution wasn't the OpenAI API.

The solution was building a system where each component supported the others.

That's why I enjoy RevOps work so much. Most organizations don't need someone to show up and sprinkle AI over every process.

They need someone to identify the bottlenecks, understand the constraints, and build a realistic roadmap.

Technology is only one piece of that puzzle.

Building systems that actually work?

That's the magic.