Want to create interactive content? It’s easy in Genially!

Get started free

7757-EN-CHAP2-INFOG2

Créa Kokoroe

Created on February 4, 2026

Start designing with a free template

Discover more than 1500 professional designs like these:

Akihabara Connectors Infographic

Essential Infographic

Practical Infographic

Akihabara Infographic

The Power of Roadmap

Artificial Intelligence in Corporate Environments

Customer Profile

Transcript

Limitations and risks to be aware of

BEFORE

launching an agent

The agent is highly dependent on the quality of the data

The agent may misuse a tool

If your data is incomplete, obsolete or badly structured, the agent will make poor decisions.

Incorrect settings may result in an e-mail being sent inappropriately, the wrong file being modified or an irreversible action being taken.

The agent doesn't have "business intelligence"

It won't spontaneously understand your priorities, your business rules or the human nuances of a professional environment.

The agent is not meant to automate everything

The agent may make mistakes in its reasoning

Automating without thinking about what role humans should play can lead to inappropriate decisions, a lack of control or a loss of responsibility.

An agent follows its own logic and may interpret a goal differently from what you had in mind.

>

Tips

The agent is highly dependent on the quality of the data:

👉 Use reliable and up-to-date sources👉 Limit access to sensitive data and check what the agent is consuming

The agent may misuse a tool

👉 Sandbox test your agent before actually activating it

The agent doesn't have "business intelligence"

👉 Clearly explain the rules, constraints and exceptions 👉 Give concrete examples of what you want from it... and what you don't

The agent is not meant to automate everything

👉 Clearly define what can be automated and what needs to be validated or decided on by humans 👉 Identify the stages of the process where there still needs to be a human in the loop: validation, decision-making, final responsibility 👉 Start with scenarios where the impact of an error will be low

The agent may make mistakes in its reasoning

👉 Start with simple, well-defined tasks 👉 Require human validation for sensitive actions

OUR TIPS