Workshop · 3 hours · online

Who decides what your AI (Artificial Intelligence) agent does when it gets it wrong?

When an agent makes a decision, handles an exception or chooses whether to involve a person, it is already shaping the customer experience. The workshop gives your engineers and developers the tools of service design to design it on purpose.

The workshop in numbers

5 questions
the team works on them, on your use case
3 hours
online, in English
4-5 people
per group
2 facilitators
UX (User Experience) and AI

1. The workshop

Designing an agent means designing behaviours, accountability and recovery.

The workshop works on a concrete situation chosen together, with two Conflux facilitators experienced in UX and in UX for AI.

Before

Writing code

The team designs features and flows. Every screen and every step is decided in advance, and the system behaves as the code says.

With agentic AI

Orchestrating autonomous actors

The team defines constraints, handoffs, checks and governance. The agent chooses the steps, and the team has to decide where it can act alone, where it stops and who is accountable for what.

2. The 5 themes

The questions your team answers before sending an agent into production.

  1. 01.

    Autonomy and limits

    Where can the agent act alone, and where must it stop?

  2. 02.

    Handoff and human-in-the-loop

    How does it hand over to a person, and how much control does that person have over the system?

  3. 03.

    Errors, exceptions and recovery

    What happens when something leaves the expected path, and how do you recover?

  4. 04.

    Accountability and constraints

    Who is accountable for what, and which limits must the agent respect?

  5. 05.

    Feedback loops and consistency

    Is the agent's behaviour consistent with the experience you want to offer your customers?

If your team has a written answer to all five, you don't need the workshop. If the answer to any of them is «it depends», those are 3 hours well spent.

3. How it works

Three steps, from the use case to material you can reuse.

First a call to choose the case, then 3 hours of work, and finally the organised material.

  1. 01.

    Before · Use case

    A call to agree on the use case. We study it in advance and prepare the online board with working templates built on the five themes.

  2. 02.

    During · Workshop

    3 hours online, for a group of 4-5 people, with two facilitators. The team works on its own agent, theme by theme, and takes written decisions.

  3. 03.

    After · Wrap-up

    We organise the material produced and deliver a PDF (Portable Document Format) handbook with the key points and the tools to reuse on other projects.

4. What the team keeps

A common language across technology, design and business.

Online board

  • The use case, chosen and studied
  • Working templates on the five themes
  • The decisions taken during the workshop, organised

PDF handbook

  • Key points of the workshop
  • Tools to use on your next agents
  • Blank templates to reuse

The decisions taken in the session stay with the team, on the board. The handbook does not contain a full analysis of your use case: going deeper is work for the team, or a follow-up project. To prepare the decisions on your agent, you can start from the Agentic Design Canvas.

5. Who it is for

When it makes sense.

Before building an agent

The team has a use case and needs to decide scope, handoffs and accountability before writing the first prompt.

When a PoC (Proof of Concept) doesn't scale

The agent works in the demo and breaks in production: the workshop finds the theme nobody has decided on.

To align development, design and product

Three teams talking about the same agent with three different vocabularies come out with shared, written decisions.

6. Book a call

In 20 minutes we'll understand whether the workshop makes sense for your team.

€3,500 + VAT (value added tax)

Starting price for one workshop: use-case preparation, board with templates, 3 hours with two facilitators, handbook.

Is your team working on AI agents?

7. FAQ (frequently asked questions)

Frequently asked questions.

Who is the workshop for?

For teams of engineers and developers building agentic systems, and for the people who work with them on product and design. No UX training is needed: we bring the service design tools.

Do we need an agent already in production?

No. You need a concrete use case, even at the design stage. We choose it together in a call before the workshop and study it in advance.

How many people can take part?

A group of 4-5 people, so you really work on the case instead of attending a lecture. For larger teams we organise several sessions.

Is the workshop online or in person?

Online, in English, on a shared board. Any in-person sessions are agreed separately, with the related travel costs.

How much does it cost?

From €3,500 + VAT per workshop, including use-case preparation, the board with templates, 3 hours of work with two facilitators and the PDF handbook. The price goes up with more sessions or more participants.

What does the team keep afterwards?

The decisions taken during the workshop, organised on the board, the blank templates to reuse on other projects and a PDF handbook with the key points.

Want to try the five questions on your agent?

A 20-minute call, with no commitment, to understand whether the workshop makes sense for your team and to choose the use case.