1. Home
  2. Packages
  3. AI Agent Pilot

Implement

AI Agent Pilot

For companies with a clearly defined use case that should run in production and deliver measurable benefit.

StageImplement
Duration6 to 8 weeks
Formatremote and on site
Your time commitmentapprox. 3 to 4 days, spread out

A good fit if …

  • A use case has been selected, e.g. from the AI Roadmap
  • Employees spend a long time searching for information or keep answering the same questions
  • A first attempt with Copilot was not convincing
  • You want to measure the benefit before scaling up

What you have at the end

  • A productive agent in your environment, e.g. Microsoft 365, Copilot Studio or Azure
  • A question catalogue against which the quality of answers is measured
  • Trained key users who can look after the agent
  • Documentation and handover

A productive agent in six to eight weeks

  1. 1Weeks 1 to 2

    Prepare

    Define goal, metrics and question catalogue, check content and permissions.

    Your contribution: Subject-matter expert with a time budget

  2. 2Weeks 3 to 5

    Build and test

    Build the agent, test it against the question catalogue and refine it.

    Your contribution: Feedback on test answers

  3. 3Weeks 6 to 8

    Pilot operation

    Operation with a test group, measurement, training and handover.

    Your contribution: Test group of up to ten people

Scope

Included

  • One use case, one agent
  • Preparation of the necessary content and sources
  • Test operation with up to ten people
  • Two training sessions for key users

Not included

  • Licences for Microsoft 365, Copilot or Azure
  • Cleansing of large document collections beyond the use case
  • Rolling out to further areas – the AI Companion takes care of that

Your contribution

  • A subject-matter expert with a fixed time budget
  • Access to the environment and the content
  • A test group from the business unit

Add-on · Change Module with Heike Baur-Wagner

So your team actually uses the solution.

New systems and AI agents only work if managers and teams embrace them. The Change Module can be added to this package.

  • Sparring for the management on the message and their role in the change
  • Workshop with the affected team to clarify questions and concerns early on
  • Support during the first weeks after go-live until usage is visible
Heike Baur-WagnerHeike Baur-WagnerOrganisational development, change, leadership and team developmentMore about the Change Module
From our projects

AI agents for business processes

As Technology Lead of an AI start-up, Frank Dolibois co-developed AI agents for business processes.

From our projects

Knowledge base with Copilot agents

Concept and blueprint for a knowledge base on SharePoint with Copilot agents in customer service.

Next stepThe AI Companion keeps the agent up to date in day-to-day operation and builds further agents with your team.To the AI Companion
Frank Dolibois

Your point of contact

Frank Dolibois

Frank Dolibois connects business processes, data and agentic AI. He implements what he recommends himself in Copilot Studio and Azure.

More about us

Frequently asked questions

What does the package cost?

The budget depends on the agreed scope and is fixed bindingly before the start. You will receive your proposal within 48 hours of the initial call. We will also clarify whether your project is eligible for funding.

Can the scope be adjusted?

Yes, in the initial call. After that, the agreed scope applies. There are no change requests within this scope.

What if the agent doesn’t reach its goals?

The question catalogue shows this early. You then know whether the content, the instructions or the use case itself is the problem – before you invest further.

Who owns the agent?

You do. It runs in your environment; configuration and documentation belong to you.

Do you have a use case for an AI agent?

In 30 minutes we clarify your situation and whether this package is the right starting point.