Back to articles

Defining Seniority for AI Agents: It's About Trust, Not Just Capability

•2 min read•By Ioannis Zempekakis
AI AgentsToqanLeadershipProduct StrategyTrust

Most questions I get about AI agents are really questions about trust.

I've been getting a lot of questions from professionals across different fields about how we define seniority for AI agents at Prosus.

The Honest Answer

It's as much a science as it is an art—and it requires lots of fine-tuning as we discover new insights.

There isn't a single metric that works on its own. What matters is how an agent behaves in real workflows, not just what it can do in theory.

Our Multidimensional Approach

We combine quantitative and qualitative signals, supported by a fine-tuned tagging system that respects privacy while still allowing meaningful differentiation.

Here's how we think about it:

🧑💼 Junior

Entry-level agents with developing capabilities.

  • Operate with a limited set of tools and integrations
  • Minimal memory and context usage
  • Primarily focused on low-complexity tasks with well-defined boundaries
  • Foundational problem-solving abilities

🏆 Medior

Mid-level agents with solid expertise.

  • Demonstrate a balanced ability to handle moderate complexity
  • Leverage tooling effectively
  • Capable of combining tools and contextual reasoning
  • Limited but purposeful use of memory

🌱 Senior

Highest level of seniority.

  • Capable of managing complex, multi-step workflows across numerous tools (+12)
  • Handle high complexity (+4)
  • Highly contextual and requiring deep domain expertise
  • Careful fine-tuning to deliver substantial value

Avoiding the Common Trap

This framing helps avoid a common trap: Treating seniority as a proxy for "how smart" an agent is.

Instead, it defines seniority by:

  • How much responsibility an agent can carry
  • How reliably it performs
  • Whether it fits into real day-to-day systems

Leadership Question

At a leadership level, this becomes a question of delegation and control, not technology.

Where have you seen unclear seniority create friction once AI became part of daily operations?

I'd love to hear your experiences and thoughts.

Originally published on LinkedIn