---
title: "Minerva - Planner Agent"
url: "https://docs.noreja.com/en/article/minerva-planner-agent"
language: "en"
last_updated: "2026-02-28T11:50:20.152Z"
---

# Minerva - Planner Agent

Minerva’s Planner Agent uses a structured “plan-then-execute” architecture to handle requests reliably—especially when questions are complex or ambiguous. Instead of generating an answer immediately, Minerva first clarifies the user’s intent, breaks the goal into logical steps, and only then executes each step using the right internal capabilities and data sources. This separation improves reliability, reduces misinterpretation, and ensures constraints such as permissions, rules, and governance requirements are considered from the start—resulting in consistent, traceable answers that mirror how expert analysts work.

**Minerva works with a so-called Planner Architecture.**  
This means Minerva does not generate answers impulsively. Instead, it follows a structured and transparent approach to reliably handle even complex requests.

### The core idea

Minerva clearly separates two steps:

-   **Planning**: understanding *what* needs to be done
-   **Execution**: carrying out *how* it is done

This approach is comparable to how a human would work: first think through the problem, then solve it step by step.

### How Minerva processes a request

At a high level, every request follows this flow:

1.  **Understanding the goal**  
    Minerva analyzes the request to identify the user’s intent  
    (for example: analysis, explanation, comparison, or recommendation).
2.  **Creating a plan**  
    The goal is broken down into logical steps.  
    Typical questions at this stage include:
3.  -   What information or data is required?
    -   Which analyses or checks make sense?
    -   Which rules or constraints apply?
4.  **Targeted execution**  
    Each step is executed in sequence using the appropriate internal capabilities or data sources.
5.  **Result synthesis**  
    All intermediate results are combined into a clear, consistent, and understandable response.

### Why this matters

For users, the Planner Architecture provides clear benefits:

-   **Higher reliability**  
    Answers are based on a defined process rather than ad-hoc generation.
-   **Better transparency**  
    Results are produced logically and step by step.
-   **Reduced misinterpretation**  
    Complex questions are systematically decomposed instead of oversimplified.
-   **Consistent handling of rules and data**  
    Constraints, permissions, and guidelines are considered from the very beginning.

### In simple terms

You can think of Minerva as a highly structured assistant:

> First deciding **what** needs to be done,  
> then carefully executing **how** to do it,  
> and only then presenting the final answer.