Minerva - Frontier Agents

Minerva Frontier Agents in Noreja – what are they?

Frontier Agents are autonomous, AI-powered agents within Noreja that continuously work in the background on a specific question or objective. Instead of reacting only to individual, manual analyses, they pursue a goal over time: they monitor developments, detect changes, validate hypotheses, and proactively deliver insights — always grounded in the process and knowledge context available in Noreja.

An important distinction: Frontier Agents are not one-off “queries”. They are more like digital team members that keep working on the topic. They continuously analyze data and models, keep an eye on defined metrics or rules, and notify you as soon as something relevant happens — for example when process performance deviates, when data sources change, or when compliance risks emerge.

Autonomous, yet controllable: started and stopped by users

While Frontier Agents operate autonomously, they remain fully user-controlled. That means:

Users can explicitly trigger agents, for example by providing a concrete question (“Why is the lead time increasing this month?”) or a monitoring task (“Continuously check whether critical compliance rules are being violated.”).

Agents can be stopped at any time when the task is completed or no longer needed.

In practice, their work can be defined through clear goals, time ranges, or triggers (e.g., “alert on deviations”), so the agent knows exactly what to focus on.

This creates a balance between autonomy in the background and transparency and control in the foreground.

Data foundation: how Frontier Agents use “knowledge” in Noreja

To ensure Frontier Agents don’t work in isolation, they rely primarily on the knowledge and data building blocks provided by Noreja. In particular, they draw from three core sources:

Event Knowledge Graph
The Event Knowledge Graph is the central foundation for process intelligence: it stores all process models. Frontier Agents use these models to understand process flows, identify variants, interpret deviations, and assess the impact of changes.

Uploaded context (documents)
Users can also provide documents as context — for example policies, work instructions, audit evidence, definition catalogs, SLAs, or process descriptions. Frontier Agents use this context to interpret analyses correctly (e.g., “What counts as a violation?” or “Which KPI definition applies?”) and to present results in your organization’s language.

Central Knowledge Graph (tool functions, features, capabilities)
The Central Knowledge Graph contains information about Noreja’s tools, features, and functionality. This enables Frontier Agents not only to produce insights, but also to explain how to implement them in Noreja — for example which analysis, feature, or modeling option best fits the solution.

With these three sources, Frontier Agents can work in a way that is process-aware, context-sensitive, and deeply integrated into the platform — rather than producing generic statements.

First Frontier Agents

1) Analyst Andy

Focus: Monitor process performance, identify root causes of deviations, propose optimization opportunities.
Analyst Andy continuously tracks the performance of your processes and detects when metrics such as lead time, waiting times, rework, cost, or SLA compliance change noticeably. But Andy doesn’t stop at highlighting deviations: he identifies likely root causes (e.g., specific variants, bottlenecks, workload spikes, or systematic loops) and derives concrete opportunities for improvement. The goal is to help teams not only react when something “breaks,” but to receive early guidance on what is changing and why.

2) Builder Benny

Focus: Suggest changes to the process model, detect schema changes, monitor data imports and their performance.
Builder Benny supports teams where processes and data evolve continuously. He suggests updates or improvements to the process model when new variants appear or when existing modeling assumptions no longer match reality. In addition, Benny helps detect schema changes in your data sources early (e.g., new fields, changed formats, renamed attributes) and monitors data imports and their performance. This reduces the risk that analyses silently degrade because the underlying data has changed — and gives teams fast, actionable guidance on what needs to be adjusted.

3) Compliance Conny

Focus: Monitor compliance topics, validate execution against context data, produce compliance reports for stakeholders.
Compliance Conny specializes in governance and compliance. She monitors defined compliance requirements and validates actual process execution against contextual information — for example policies, audit criteria, or mandatory process steps described in uploaded documents. Conny flags anomalies (e.g., missing approvals, incorrect sequences, unauthorized variants) and generates compliance reports for stakeholders — clear, traceable, audit-friendly, and written in the context of your organization.