The video

The presentation explored possible uses of AI agents in customer service, sales and analytics. A central recommendation was to begin with one bounded task and describe which data the agent may read, which actions it may take and how its output will be checked.

What this means for the organisation

A sound first trial has a measurable task, a known consequence of error and an accountable process owner. Before scaling, the organisation should compare quality, time and exceptions with the current workflow and adjust permissions accordingly.

The video covered agents in service, sales, analytics and reporting, while recommending one clearly defined first task. This limits both data access and possible faulty actions. Only after quality and exceptions are measured against the current process is there a basis for expanding an agent’s sources or actions.

NIST AI RMF structures risk management around govern, map, measure and manage. It provides a vendor-independent frame for reading ProsessPilotene’s guidance and model tests. The demonstrations show particular tasks; the framework also requires organisations to assess context, consequences, measurement and responsibility throughout the lifecycle.

A pilot needs a bounded dataset, expected answers, difficult counterexamples and a named decision owner. Errors should be classified by consequence rather than merely counted. Model and prompt versions should be retained so the organisation can explain changed results before giving an agent wider scope to act.

Sources

Primary source: ProsessPilotene, “Konkurransefortrinn med AI-agenter,” 1 March 2025 (Norwegian).

NIST, AI Risk Management Framework 1.0, 26 January 2023.

Join the discussion

What would you investigate first before using this? Share your questions or experience in the discussion.