Introduction
Manychat has grown from a small founding team in 2016 to more than 400 employees across 24 countries. Dmitry Kushnikov, the company's first employee and now both CTO and COO, has guided Manychat through remote work, office life, hybrid models, and global expansion. In a recent discussion, he explained how artificial intelligence is reshaping hiring, team structure, and performance evaluation at the company.
What Happened
Kushnikov described a more disciplined approach to hiring. Before opening a new role, Manychat leaders ask whether the work needs to exist at all and whether it can be automated. Because AI agents can now handle a wider range of tasks, the bar for adding headcount is higher than it used to be.
He also revisited the company's 2018 shift to cross-functional teams under Large-Scale Scrum. The change addressed an accountability problem: in functional silos such as backend, frontend, and design, it was unclear who owned final results. Today, teams own customer problems end to end. Kushnikov noted that the model is being reexamined because AI lets small teams accomplish far more than before.
On performance, Manychat recently published "The Manychat Way," ten operating principles built from five long-standing values. A key principle is that the company does not track LLM usage or token counts. Instead, it celebrates outcomes. Managers agree on measurable results before work begins, avoiding the trap where a metric becomes the target and stops reflecting reality.
Customer support illustrates the approach. After ticket volume began outpacing the team in early 2024, Manychat placed an AI agent on the front line in 2025. Since the end of 2024, monthly ticket volume has roughly doubled while the frontline team grew by only about a third. Customer satisfaction rose from roughly 84 percent at the start of 2024 to about 90 percent through 2025.
Why This Matters
Many organizations are rushing to measure AI adoption by counting prompts or tool usage. Manychat is taking the opposite path. Kushnikov argues that activity metrics can distort behavior, especially when people optimize for the number rather than the result. By separating how work is organized from who reports to whom, and by keeping humans in the loop for sensitive areas such as payments, security, and data access, the company is trying to adopt AI without losing accountability.
The implications reach beyond one company. As AI agents handle routine work, the manager's role is shifting toward judgment, vision, and responsibility for people. Kushnikov calls this the "visionary operator" model, in which leaders complement technology instead of competing with it.
Key Takeaways
- AI agents raise the threshold for creating new roles by making automation a realistic alternative.
- Cross-functional teams at Manychat own customer problems end to end, improving accountability.
- The company does not measure employees by AI tool usage; it measures outcomes.
- Customer satisfaction improved as AI handled first-line support, even as ticket volume grew.
- Leaders are expected to adopt AI themselves before asking teams to do the same.
Conclusion
Manychat's experience suggests that AI works best when it complements human judgment instead of replacing it. By focusing on results rather than activity, and by giving every employee access to AI tools while documenting processes in a shared repository, the company is preparing for a future where small teams can do more. Whether other organizations follow a similar path remains to be seen, but the emphasis on outcomes over outputs offers a useful counterpoint to the rush to quantify AI adoption.

Discussion
Join the conversation
Thoughtful reactions, questions, and follow-up ideas help shape the next story.