worksmarto · Multi-Agent LearningSample session

Blog Squad

What you'll learn

How a team of AI agents talks, disagrees, votes and approves a short blog post, and who has the authority at each step.

Level: Beginner · Stack: LangGraph.js, Groq, Gemini, OpenRouter

The 30-second take

  • Give a topic, or let the team pick a trending one. After that the agents do everything on their own: no human in the loop.
  • Press Play at the round table: each message pops up, lines show who it's for, tags show each seat's power.
  • The side panel shows the Plan (workflow and who approves what), the task board, Draft 1 and Draft 2, the chat, approvals and the Final post.
  • Stuck? Ask the Coach, bottom right. The Learner's Guide at the bottom explains how it's built.
  1. 1TopicYours, or a trending pick
  2. 2PlanTeam Lead sets tasks and the bar
  3. 3Vote3 agents vote, title locked
  4. 4Draft ⇄ reviewEditor scores, Writer revises
  5. 5ApproveTeam Lead signs off

1. Give the team a topic

Topic
Length
Tone
Optional teammates The Team Lead decides whether to call each one.
Translates the final post
Connecting…

2. The round table

Who speaks, who listens, who decides
Now: –

3. The drafting desk

Every draft side by side. Blue shows what the Writer added, crossed-out orange shows what was cut. Under each draft: the Editor's notes, the Writer's answer and the Team Lead's ruling.

Added in this draftCut in the next draft

4. The finished post

Learner's guide

5. How this multi-agent team works

Everything you just watched is pre-coded: each agent has a fixed job, fixed inputs and outputs, and the team follows fixed rules for talking and approving. Here is how.