(NewsNation) — As 911 call centers across America struggle to recruit and retain enough staff to meet their volume of calls, one center in South Carolina is using artificial intelligence to help ease the volume of calls.

This year, the Charleston County emergency call center started using AI to redirect some non-emergency calls to an automated system, easing the burden on dispatchers who also handle 911 calls.

“We in Charleston County are suffering from staffing shortages similar to most of the other agencies around us in the country,” said James Lake, the Director at Charleston County Consolidated Dispatch Center. “And what we’ve decided to do is not only are we focusing on our recruitment and retention of people we do, we’re taking another look at our workload and how we can better manage that.”

So authorities connected the call center to AmazonConnect, an AI-powered software that helps automate customer calls.

The goal is to filter out calls that don’t require immediate attention from dispatchers — what Lake called “administrative calls.” Most of those calls are for animal control, Lake said.

“These calls are about feral cats, rabbits in the backyard, raccoons, you know the animal control calls especially they’re not about calls that we send responders immediately to the scene,” he said.

The AmazonConnect system is in place for the county’s nonemergency line with the goal of easing the call volume to the center, allowing dispatchers to focus on emergency calls.

When someone calls the non-emergency line, they are prompted to offer a description of the service they’re requesting.

For example, if the person says “animal control,” they’re then asked if the animal is immediately threatening someone. If they aren’t, they’re offered the option to receive a text link to an online reporting system where they can file a report. Also, if the system fails to hear you, it will transfer you to a dispatcher.

The county is spending an estimated $3,000 a month on the technology. In the first month, Lake said the AI-assisted automated system handled 36% of the non-emergency calls.

“We saw an immediate difference, and we haven’t done a strong marketing effort yet,” he said.

He also shared figures showing that there was a reduction in administrative calls that ended up being taken by operators.

The number of administrative calls dropped from 16,775 last June compared to 11,201 calls last month.

Lake stressed that the system is not a replacement for human dispatchers. He said they got about 300,000 calls last year, meaning the need for dispatchers is still urgent. As of April, they had just 67 call takers and dispatchers when their goal was to have 124.

But Lake still thinks the new system is helping ease the load of calls. His advice to other departments who are considering this type of system is to take it slow and steady.

“First of all, identify where you can realize efficiencies and then if you’re implementing a tool like this, start off slowly. Don’t try to solve all of the world’s problems at once, take it slowly piece by piece and learn from what you’re doing,” he said.