At Omnisend, employees can earn a permanent raise by finding ways for AI to take over parts of their work. The email and text-message marketing company offers an additional 2% to 4% salary increase for demonstrated business impact from AI adoption, on top of regular raises. It says 63% of eligible employees earned the increase in the first quarter.
The arrangement gives workers a reason to experiment, while changing the economics of growing the company.
“We continue to increase salaries and offer additional raises for AI adoption, but we’re hiring less,” Rytis Lauris, Omnisend’s co-founder and CEO, tells me in an interview. “We’re bootstrapped, or what we call customer-funded, so we don’t have external capital. Under our operating model, whenever we had spare cash, we hired more people. Headcount grew steadily throughout the year. That growth has slowed because our existing employees are becoming increasingly efficient.”
Headquartered in Vilnius, Lithuania, Omnisend has roughly 260 employees and serves more than 150,000 clients. Its voluntary salary program offers a concrete incentive for workers facing an uncomfortable proposition: helping AI do more could change how much human labor their employer needs. Can paying employees to lead that transition give them a meaningful stake in its benefits?
Why AI Adoption Leads to Salary Bump
Lauris built the AI incentive into Omnisend’s existing compensation process, which has included quarterly salary reviews for roughly eight years. The additional raise rewards employees for developing skills the company sees as essential to its long-term growth.
“We fundamentally believe that if we want to build a future-proof organization, we need employees who are ready for that future. That means learning how to work with this new technology, which is fundamentally changing how we do our jobs,” Lauris says. “If you want to remain relevant in five or 10 years, you have to start investing now. A factory invests in new machines. In our business, the products we build come from human knowledge and expertise. That is why we invest in developing our employees’ skills and helping them change how they work.”
Bernard Meyer, Omnisend’s head of AI operations, says employees qualify through useful work that colleagues can adopt. Recipients extend beyond technical teams. His concern is whether an experiment changes performance enough to justify continuing it.
“For nearly four quarters, we’ve focused explicitly on outcomes. We don’t want employees simply to stay busy using AI. We want them to use it in ways that create value and help them achieve the key results they set at the beginning of each quarter as efficiently as possible,” Meyer says. “At the end of the quarter, we measure their progress against those results. We want every department across the organization to focus on how AI can help it achieve its goals faster and make the company more productive, ultimately delivering a better experience for our customers.”
An internal initiative called “AI as a habit” supports that effort. Meyer says asking employees to adopt AI means helping them figure out where it belongs in their work.
“I’ve seen many examples of senior leaders or middle managers telling their teams, ‘We need to use AI, we need to use AI.’ But that is where the conversation begins and ends,” he says. “They don’t build the infrastructure to support that expectation or make the investment to help employees put it into practice. People then say, ‘I know I need to use AI, but I don’t know what I’m supposed to do with it.’”
Engineers Stop Being the Bottleneck
Omnisend’s finance team aimed to automate 36% of its invoice-processing workflow and reached 85%, according to the company. That work includes assigning cost centers and routing invoices into accounting systems. Meyer makes a distinction that matters when compensation depends on AI. Employees can earn recognition for using it to build conventional software, without requiring the finished system to keep asking a model what to do.
“Some of the tools our teams build don’t actually use AI in the finished product, but AI is what enables those teams to build them in the first place,” he says. “When we assess AI usage across Omnisend, we recognize both approaches. For tasks that follow clear rules and produce predictable results, you generally don’t need AI in the workflow itself. Conventional automation can be the better fit.”
Other initiatives offer financial measures. Omnisend says an automated payment-recovery workflow recovered 92% of failed-payment revenue in one quarter, representing about $324,000 in monthly recurring revenue. Its automated business development representative, Mary, responds to signup events in HubSpot. The agent gathers context before choosing an outreach channel and message, then sends qualified leads to an account executive. The system grades interactions and tracks outcomes, including revenue attribution. The company credits Mary with sourcing $27,755 in new-business revenue to date.
For customer retention, product and engineering employees choose reported problems worth fixing and use the AI development tool Devin to resolve them, shortening the usual planning and handoffs. The company’s latest update attributes $580,945 in monthly recurring revenue protected from churn to that effort.
Lauris says AI also lets product managers and designers resolve smaller problems without waiting for engineers. Customers increasingly expect similar flexibility. He describes Omnisend’s connections to Claude and ChatGPT through the Model Context Protocol as among its fastest-adopted features.
Build the Agent, Own the Consequences
As employees gain the ability to build software, responsibility for its behavior becomes harder to separate from their everyday jobs. Meyer says ownership generally stays with the team that starts a project. “The teams that initiate a project retain ownership of it afterward. There isn’t a separate team stepping in to say, ‘This is what you should build, and we’ll take ownership once you’ve built it,’” he says. “We also keep permissions limited, starting with read-only access rather than allowing the tool to make changes. When we do grant permission to make changes, we first test that capability in a testing environment.”
Teams test unusual scenarios before introducing small pilot groups. The legal department built its own AI tool to assess proposed software, allowing employees to proceed after a green result and referring uncertain cases for further review. It also uses automated initial contract checks that humans review. Omnisend’s E-hawk screening checks new signups for abuse. When accounts pass screening but later receive bans, the company examines the patterns and develops new detection rules. The process acknowledges that approving a system once does not settle how it should behave indefinitely.
Lauris also describes an experiment whose initial scope proved too ambitious. Customer success managers tried to build a comprehensive AI assistant that would analyze account performance and help them prepare for client conversations. “They were building a comprehensive assistant to reduce the time they spent preparing for customer conversations. Before each call, an account manager reviews the customer’s account to understand how it has performed,” he says. “The tool didn’t work as intended, even though the team had already invested considerable time in building it.”
The team reduced the project’s scope. A narrower tool now creates standardized quarterly business review presentations. Lauris says it saves substantial time, although less than the original plan promised. He did not clarify whether the unsuccessful work earned employees a raise.
What Happens to Employees Who Opt Out?
Participation remains voluntary, but Lauris sees adapting to AI as essential. When I asked how Omnisend keeps the program fair for employees who decline to participate, he pointed to managers’ discretion. Employee surveys and feedback help identify dissatisfaction, he says.
That leaves managers with considerable influence over who earns more. Lauris’s expectations nevertheless extend across the company. “Every individual contributor is becoming a manager in the sense that each of us will need to put AI agents to work to become more effective and efficient than we were before,” he says. Lauris expects familiar job titles to remain even as the work behind them changes. “Your position may still have exactly the same title, but the way you carry out your day-to-day responsibilities will change significantly,” he says.
The Control Layer Verdict
Omnisend gives employees a financial stake in automation and greater control over how their work gets done. That freedom brings responsibility for software they once needed specialists to build, while managers decide which contributions deserve a permanent raise. Slower hiring suggests the arrangement can reward existing workers even as growth creates fewer openings for others. Employee-led AI adoption still needs clear rules about authority and reward. Omnisend’s model will become more convincing if employees can see how the company values their judgment as clearly as it measures the work their agents complete.
