Zembl is an Australian energy consultant that helps businesses of all sizes compare and switch energy plans, working with a panel of leading retailers across the country. When inbound lead volumes started climbing and a persistent after-hours problem began costing the team real deals, they faced a familiar fork: hire more people, or build a smarter system. They chose to build — and ended up with something neither option had promised: an AI agent workforce that now operates across their sales, marketing, and operations teams around the clock.
The Challenge
Zembl’s inbound sales model depended on fast, personalized contact with every prospect. But before a rep could make a meaningful first call, they had to do the groundwork — researching the business on Google and LinkedIn, validating contact details, tracking down an energy bill. Multiplied across every new lead, it added up fast.
“We have highly qualified sales people here to sell to businesses and them wasting their time on low value tasks was a bit of a pain,” said Edan Shelley, Head of Acquisition and Sales Enablement.
Speed was one issue. Consistency was another. Some reps were rigorous with their research; others weren’t. “You would often have inconsistencies in the way that data was either gathered or managed and then stored,” Edan noted. When a customer submitted a single contact detail and a rep didn’t dig further, that was a missed number, a missed conversation, and a missed deal.
Then there was the after-hours gap. Prospects inquired late at night and on weekends — but Zembl’s sales team worked business hours. “If you were to opt in on a public holiday or on a Saturday morning, you are apart from a sort of a very generic email confirming your expression of interest, you wouldn’t speak to one of the sales team till maybe midday on a Monday,” said Edan. By then, the lead had often already gone elsewhere.
As volumes grew, the team mapped out a solution. “Edan and myself, we wanted to hire a junior person to help with pre-qualifying leads on the inbound team. We put a JD together in a business case,” recalled Sunny Singh, Head of Digital Marketing. Then AI agents entered the picture.
The Solution
A referral from a peer at a fellow Australian tech company led Sunny and Zembl’s Director of Sales to Relevance AI. What they found wasn’t just another automation tool.
“It just felt like we had control in that platform to do stuff and set some guardrails and it was, you know, enterprise ready. And it just wasn’t another tool. It was something that was going to help us build like a workforce,” said Sunny.
They started building in October 2024. Their first agent — Toni — went live in January 2025, and was onboarded like a new hire. Every email Toni sent was reviewed before it reached a prospect. The team started with five leads a day, then ten, then fifteen, building confidence in the output before removing the guardrails. “We were approving everything, like every email that was going out to all the leads at first, uh, we would feed Toni like for the first week, five leads a day, then the following week it was 10 leads and there was 15 leads. We were just to get comfortable with what Toni was saying,” said Sunny.
When Toni eventually got overwhelmed by the volume and scope of tasks, the team didn’t scale back — they gave Toni a direct report. Salesforce Sally joined to handle all Salesforce-related work: uploading bills, updating fields, routing leads. SMS Sammy came next, triggered whenever an email bounced. Then Comparison Cami, running high-level energy comparisons in the background as soon as a bill came in. Toni now has three direct reports.
The team also applied the same performance processes to their AI agents as they do to every Zemblr. “We did a 360 review with Toni. So just like you would do with humans, we did the same thing with Toni,” said Sunny. Those sessions — run as whiteboard meetings with the whole inbound team — created a development roadmap for Toni and brought the broader team along with it.
Today, Zembl has around 10–11 active agents running across the business. Newsletter Nelly sends a weekly energy news roundup to the team every Monday. Ranky updates hundreds of SEO pages for LLM optimization. Dory monitors contract renewals and flags changes in key account details to opportunity owners before conversations happen.
The Results
The impact from Toni was visible within weeks. The rate of customers uploading an energy bill before their first sales call almost doubled in the first six weeks of Toni going live.
“Within the first six weeks of Toni up and running, that had almost doubled. So we’d seen twice the amount of customers before they’d hit the floor upload an invoice,” said Edan. The knock-on effect on the first call was direct. “Instead of a conversation being, hi, could you provide me your invoice? We’ll run a review. It was, hi, we got your invoice. You spoke to my colleague Toni. The result is X savings. And it made for a much more fluid process and customer journey.”
With leads pre-researched, personalized, and bill-ready before a rep picked up the phone, average call time dropped significantly. “I think it’s from 22 minutes down to like nine now, eight minutes or something like the time saved,” said Sunny. With the same inbound team — no headcount additions — conversion climbed.
“We haven’t increased the team size in terms of the BAU team size, but the impact that’s had on the overall time saved, we’ve seen a 30% increase in overall customer conversion. So that’s 30% more customers resulting in a close one journey because we’ve been able to speak to more customers because we’re not wasting as much time on a lower value task,” said Edan Shelley.
The after-hours gap is now a non-issue. The team cite one early morning lead as a landmark moment: a prospect inquired at 5:30AM with three business sites needing review. By 8:30AM, Toni had researched all three, made contact, and collected the bills. All three sites were closed before the sales floor had even opened.
In Their Own Words
“I’d never go back to manually researching a business, finding out details of the customer’s address, phone numbers, who else is responsible for the energy contracts. The thought of doing that manually now, it’s just not a process we wanna go back to.”
— Edan Shelley, Head of Acquisition and Sales Enablement, Zembl