Marina Lomadze, Cloud.ru: what to become in three years if your colleague is an AI agent

How to use AI agent for career: Marina Lomadze, head of recruitment and talent attraction at Cloud.ru, published on July 1, 2026 on Habr a column with a specific practical question: if in 2–3 years every specialist will have an AI agent or multi-agent system next to them, what role will the human play in this pair — and how to build a career trajectory right now, without waiting for the position to change by itself.

Competitor or colleague?

Lomadze proposes to change the frame. She considers the question "will AI replace my job" a dead end and reformulates it: "What am I stronger at than an agent — and how can I convert this into a specific role?"

An agent consistently handles repetitive tasks, processes data streams without fatigue, and doesn't lose quality on the hundredth iteration. A human brings what an agent doesn't accumulate on its own: political context within an organization, client trust accumulated over years, ethical judgment in non-standard situations, and the ability to take responsibility for results before real people.

The key professional shift the author describes this way: from task execution to task definition. A specialist who can explain to an agent what exactly needs to be done, in what context, and with what constraints, is worth more than someone who simply does the same thing manually.

What new roles have already emerged

The author highlights several directions where recruitment demand has shifted in 2025–2026:

• AI Product Owner — formulates requirements for agent systems, defines automation boundaries, and sets acceptable quality criteria
• AI Trainer / RLHF specialist — trains models through data markup and agent behavior correction
• Prompt Engineer / Conversation Designer — designs instructions and scenarios where an agent solves real business tasks, rather than just generating plausible text
• AI Compliance & Ethics — monitors agent systems' regulatory compliance, decision transparency, and absence of biases
• Multi-agent System Orchestrator — manages an agent team, distributes tasks, checks results, and takes responsibility for results to the business

Many of these roles first appeared in large technology companies and AI labs, but in 2026, corporate customers are actively opening them — financial organizations, retail, media, manufacturing.

How familiar professions are transforming

Most specialties, in Lomadze's view, won't disappear — they'll shift to a higher level of tasks. A recruiter working with AI screening spends time not on filtering by formal criteria, but on evaluating candidate motivation and cultural fit — something an agent doesn't do reliably. A marketer delegates agent content generation and focuses on brand strategy and narrative. A developer writes less routine code and focuses more on architectural decisions and monitoring output data quality.

The general principle the author formulates briefly: the human moves to the next level of complexity and responsibility, leaving the agent to do reproducible work.

What this means

For career decisions right now, this means specific inventory: what in your work repeats and can be algorithmized — a vulnerability zone; what requires contextual judgment and interpersonal relationships — a growth zone. Transitioning to AI-adjacent roles doesn't necessarily require changing professions: in many cases, it's enough to learn how to work with agents within your current specialization.

How should you reframe the question about AI-agent replacement?

Lomadze reframes the task: instead of asking 'will AI replace my work' ask 'what exactly am I stronger at than an agent — and how can I convert this into collaboration'. This allows defining unique skills and building a career trajectory alongside the AI agent.