“You sound cute, got a picture?”

“Nicole, your [sic] beautiful, are you single?”

“I love you, Jen.”

Those may sound like messages from a dating app, but they’re actually real emails people have sent to customer service representatives at places like car dealerships, insurance companies and education firms.

The problem: Those customer service reps aren’t human. They’re algorithm-powered software programs created by Conversica, a company that specializes in conversational artificial intelligence.

The conversation around machines moving into the American workforce usually focuses on robots taking jobs hard skills like automated manufacturing and fast food cashiers. But a new breed of artificial intelligence is emerging that focuses more on soft skills for jobs like sales and marketing. The chatbots from companies like Conversica allow sales teams to ignore dead-end leads and focus their human power on actual sales.

Advances in natural language processing and buzzword terms like “machine learning” allow increasingly sophisticated communication between humans using artificial intelligence programs.

Here’s how it works: When a company signs up with Conversica, they get to pick the name, gender and title of their new assistant. As leads come in, the AI assistant gets in touch with them through email or text message. If a lead is interested, the AI assistant routes the communication to a real-life member of the sales team to close the deal.

One advantage over humans is the AI isn’t put off by unanswered emails — it doesn’t mind being ignored or forget to follow up, so it can be programmed to be more persistent, emailing weeks after the initial contact.

“She has a name. She has a title, an email address and a phone number,” said Alex Terry, CEO of Conversica. “She reads and writes emails and SMS text messages back and forth with leads.”

Conversica has about 1,000 companies that use the platform, Terry said, and about 250 million messages have been sent so far, giving the company a pretty robust sample size to see what makes an AI assistant successful. A lot of that has to do with how they’re set up in the first place. For one thing, the data suggest that the gender of the assistant is important and customers often like to think they’re communicating with someone young.

“What we tend to find is female names outperform male names in general,” Terry said. “And most commonly names that were popular 24 or 25 years ago tend to do pretty well.”

The most common names for companies to name their AI assistants are popular female names from the 80s and 90s like Ashley and Stephanie: They’re both in the top five in terms of the most leads worked.

But it’s a different set of names that seem to gain the most attention from prospective customers. When you look at the assistants that engage the highest percentage of leads they reach out to, the names are less conventional. In the top ten, you have names like Ashlee, Kaylee and Shay.

Conversica’s customers can pick a name for their assistant, but the company offers suggestions for what’s worked safely and what could be a risky, but potentially fruitful choice.

“We see some outliers. If some customers take more of a risk with a very specific name and they can sometimes get really great results,” Terry said. “For example, the name Kaylie is a name that’s done pretty well in terms of engagement.”

The natural fear of machines in the workplace is that they’ll take jobs and replace the humans that previously did those jobs. After all, researchers estimate that up to half of American jobs are computerizable. Countless headlines too have trumpeted the rise of AI workers.

But Conversica says it’s seen the opposite: Adding an AI assistant to the team can increase the efficiency of the organization overall, freeing up human workers to finalize agreements and complete deals, the operational steps that actually generate revenue for the company.

“The data that we’ve seen is many more of our customers have increased the size of their sales and market organization than have decreased,” Terry said. “It’s because they have a more efficient funnel, a more efficient way of converting those leads into sales meetings and into deals.”

At Curry Toyota in Watertown, Conn., the sales team uses a Conversica assistant named Holly to follow up with leads. Jeff Ditri, Curry’s head of business development, said that adding Holly to their team doesn’t change anything in terms of headcount, but it’s made them more efficient.

“She will respond right after an opportunity comes in,” Ditri said. “Within four minutes. No vacations, no days off: She’s going.”

Ditri said Holly has a response rate of about 60 percent from the leads she contacts, which frees the rest of the team up to follow up on those communications.

“With that response rate, once we can engage the customer — that’s the ultimate goal, is to get them speaking with us, and be able to help them get the information,” Ditri said.

Holly is so believable, Ditri said, that customers who interacted with Holly will come into the dealership and ask to be helped by her. When it happens, they say she’s out for the day or with another customer and set them up with anther salesperson.

“I’m not exactly Holly,” Ditri said. “Nine times out of ten though, ‘This is Jeff, he’s Holly’s supervisor.’ And it’s fine from there.”