Deep dive · module 87 of 100
Multi-agent systems
Understand how multiple agents can work together.
After this lesson
You can describe a simple multi-agent idea.
In view · Working together towards one result
Collaborating with clear handover
Collects relevant information and provides sources.
Creates a draft based on the information provided.
Compares draft and sources and reports what is still incorrect.
Part 1 of 3 · Researcher
Your preference for less movement is on. Choose each part yourself.
What is multi-agent?
Multiple specialised agents work together on a task.
Example
One agent searches for information, another writes, a third checks.
Challenge
The agents must be well-coordinated. Otherwise, confusion or duplicate work arises.
New practice example
From explanation to a usable result.
Multiple agents can divide work, but require clear handovers and a single responsible final check.
Fictional practice situationA fictitious researcher agent finds two source passages. A writer creates a text. A controller notices that one passage contains an exception that the writer omitted.
View a possible solution
This is how it can be done
Provide the writer with the passages including source labels. Have the controller verify claims against those same sources and return the discrepancy. A human chooses the final version. The fact that multiple agents repeat something is not independent proof.
Why this works
Without shared source control, agents can reinforce each other's assumptions. The handover must preserve uncertainties.
Make it slightly more difficult
Let the researcher fail to find one source. Describe how that missing evidence remains visible to the writer, controller, and final decision-maker.
Extra practice, at your own pace. You do not need to submit anything. This does not count towards an automatic assessment.
Your workplace · learning by doing
Make something. See what changes.
Checking the AI connection… You can already view the written example.
Present your assignment to Ende.
Read the answer, change one thing, and compare.
Previously held
New version
You remain in control
What do you want to improve in your next attempt?
On this device only. AI answers remain only in this page memory. This is self-review, not a grade.
Why this approach works
Without shared source control, agents can reinforce each other's assumptions. The handover must preserve uncertainties.
A step further
Let the researcher fail to find one source. Describe how that missing evidence remains visible to the writer, controller, and final decision-maker.
Try & discover · no obligation
Put the steps in a workable order
Move the steps with the arrows. Check whether each step has its necessary input and control.
What does a step depend on?
- Researcher provides sources must precede Writer creates a draft
- Writer creates a draft must precede Controller compares source and draft
- Controller compares source and draft must precede Human decides on use
Move a step and then check the sequence. The arrows also work with your keyboard.
This model checks dependencies, not duration, capacity, or actual execution. With multiple agents, the human decision remains a separate step.
This trial works locally with fictitious data and does not count towards your lesson completion. It starts anew with a new lesson or reload.
A prompt to get you started
Describe a simple multi-agent system for preparing a report. Give each agent a clear role.
Apply it
Three agents confirm the same conclusion, but all three received only one erroneous summary as a source.
Which approach best fits this situation?
- Check the original source and make assessments sufficiently independent.
- Count three agents as three independent sources.
- Only increase the number of agents without changing the input.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
What is a risk of multi-agent systems?
- Agents may contradict each other or perform duplicate work if they are not well-coordinated.
- It always works perfectly.
- It is simpler than a single agent.
Extra practice
Outline a multi-agent idea for one of your tasks, using a maximum of three agents.
Agent 1: searching for sources. Agent 2: summarising. Agent 3: checking for accuracy.
