How Chalkie grew revenue 6x without hiring a single support agent

Teachers lose time every week outside the classroom building lesson material and adapting it for students with different needs. Chalkie makes classroom-ready, curriculum-aligned resources that save the average teacher four to five hours a week. More than a million teachers in over 100 countries now use it.

At the start of 2026, Chalkie was just three co-founders, and teacher questions landed in their personal inboxes. Getting back to everyone in good time was difficult, and every reply meant stepping away from building the product. “Support was handled manually by the founders over email,” says Joe Turner, Chief of Staff at Chalkie. “There was no measurement other than getting the inbox down to zero as easily as possible.”

Then, in six months, revenue climbed sixfold to nearly $6 million, and the team grew from three to 11. Teachers around the world were asking questions at all hours, now in more than 40 languages. The obvious next hire was a full-time support person, but Chalkie decided to go another way.

Through the Early Stage Program, Chalkie brought in Intercom as its helpdesk, a platform Joe knew from a previous startup. It was quick to set up, a small team could manage it without specialists, and every teacher request now landed in one connected system. With that foundation in place, the next question was how to scale support: keep hiring as the business grew, or use AI to build a more efficient model from the start. They chose Fin, an AI Customer Agent, to help keep the cost of support from rising in step with the business. “It’s quicker to scale and optimize Fin than it is to scale up people,” says Joe.

Making that model work meant giving Fin the right knowledge first. Phil, Chalkie’s CEO, filmed walkthroughs of every feature, transcribed them, and turned them into Fin’s first articles. The team also connected Fin to Chalkie’s website, giving it access to more than 3,000 pages of existing information.

With that knowledge foundation in place, Emma O’Carroll, an operations associate at Chalkie, began pushing Fin beyond straightforward answers and into the complex, multi-step work that had previously needed a person. Refunds were the first advanced flow the team automated. They wrote out the refund policy in plain language, as though briefing a new teammate, and paired it with rules Fin had to follow exactly. Fin now asks whether a subscription is new or renewing, checks the date the customer provides against the policy, and explains where they stand before the conversation reaches a human. “Fin is doing that pre-qualification before I even have to look,” says Emma.

In the first six months after implementing Fin, Chalkie’s monthly support volume grew 27x: from 153 conversations in December 2025 to more than 4,000 by May 2026. But the team didn’t need to hire to keep up. Today, Chalkie’s entire support team is Emma and Fin.

Fin is now involved in 99% of conversations, with around 87% of Chalkie’s overall support volume resolved without a human stepping in. What's left is the work Chalkie deliberately keeps with a person: approving refunds, making account changes, and handling anything touching a customer’s payment.

That performance has held as Chalkie has scaled. From January to July 2026, monthly volume grew sevenfold, yet Fin’s resolution rate never fell below 87% in a single month. And quality improved alongside it. CX Score, an AI-powered measure of answer quality across conversations, rose from 60% in January to 76% in July.

For Emma, that has transformed support from an all-day reactive workload into more strategic, proactive work. Fin saves her around six hours a day, which she now spends building Chalkie’s community and curriculum and automating its sales process. "Fin shields us from the barrage of mundane requests you typically get in customer service," says Joe. “It's no longer a pain point for us. That's the biggest selling point: there's no noise."

For a team of 11 serving more than a million teachers, that is what made the growth possible. "You have to get ahead of the problem before it becomes one," Joe adds. "As a startup, you can't afford to waste valuable headcount on repetitive tasks."