The CEO of an AI startup says his company no longer has engineers. Instead, they became engineering managers who managed AI agents.
Tatyana Mamut, CEO and co-founder of AI agent startup Wayfound.ai, says her team of two engineers is more productive than a large engineering team. Mamut, a former director at Amazon Web Services where he managed an engineering team of more than 30 people, says two engineers “allow us to ship code much faster.”
“Two engineers shipped more features in 2017 than my team at Amazon,” Mamut said.
Today, coding agents like Claude Code do the work of engineers, and humans manage the work of engineers. The two-person team, including the CTO, also spends time interacting with customers to build custom features.
Mamut chatted with Business Insider about how the startup did this and what Wayfound got out of it.
Wayfound turned engineers into managers
Wayfound began this transition in 2024. The company’s engineers initially used ChatGPT to assist with coding. As other coding tools such as Claude Code, Vercel, and Cursor were released, engineers began testing them. Currently I mainly use Claude code.
“At some point a few months ago, Claude Cord got really good,” Mamut said.
She says the engineering team has achieved significant efficiencies. There is no need to do “weeks of testing.” It is currently being tested by a coding agent. We also use Wayfound’s internal agents to monitor code quality and suggest improvements to engineers.
When Mamut previously managed large teams, this process took several weeks. Engineering managers plan project requirements and priorities, allocate work, control quality, and collaborate with product managers. During this process, engineers may get into arguments with each other.
“We now have meetings on Mondays and Wednesdays to discuss what needs to be shipped,” Mamut said. “What are the customer’s priorities? That takes 20 to 30 minutes. They prioritize what we need to do. The engineer takes that and goes to Claude Code and says, ‘This is what we need to do, this is what we need to accomplish.’
Coding agents also helped reduce costs for startups. Wayfound didn’t need to hire additional customer success employees because their engineers interacted with customers.
“Engineers are now talking much more with customers and staring at screens much less,” Mamut said.
Mammut describes the role of these employees as “builders” rather than engineers. This title is becoming more common with companies like Meta.
“Product, design, and engineering capabilities are converging into one function,” Mamut says. “Product managers talk to customers, designers design features, and engineers code features. Now one person can do all of that.”
Mammut warned that “agent slop” can become a major problem when organizations use AI tools and agents without ongoing training and supervision by both humans and professional AI agents. He said companies deploying agents and then withdrawing could cause “big problems.”
“It’s easier than ever to start prototyping new AI agents across your organization, but few companies are able to actually execute and drive ROI without consistent oversight and refinement,” said Mamut.
Agents are the future of SaaS
Mamut also had a lot of thoughts about the traditional Software-as-a-Service business model. Companies like Salesforce, Atlassian, and Workday have seen their stock prices hurt as investors worry that AI tools will destroy or replace their software.
Mamut said the SaaS model needs to change completely. To adapt, these companies should become “agents,” she says.
“Board members are demanding that SaaS companies reduce costs to invest in AI,” Mamut said. “The only way we can reduce costs is by reducing the number of employees, so we are seeing a significant reduction in headcount and employment.”
Winning companies will add AI agents to their products, but if they try to maintain the core SaaS model, “it will disappear within five years,” Mamut said. This means you need to invest money, time, and AI tokens to build these agents. Profits may take a hit now, but it will pay off later.
As companies look to reduce costs and headcount, they don’t want to be locked into multi-year contracts that are a key part of traditional SaaS models. Because we don’t even know how many seats we’ll need in the future.
“This is a change that most SaaS companies need to make,” Mamut said.
While AI agents can perform some tasks traditionally performed by humans, Mammut sees humans as agent managers, customer relationship builders, and people who set the company’s strategy.
“People want to buy from other people,” Mamut says. “People want to know who they can trust. You need to build those relationships.”
Any tips? To contact this reporter via email, please specify the following address: rmchan@businessinsider.com Or send a signal at rosal.13. Use a personal email address, non-work WiFi network, and non-work device. Here’s a guide to sharing your information securely.