How I Built an AI Research Team in Moxt (Step-by-Step)

I built a 5-agent AI research team in Moxt that turns one sentence into a fully cited, illustrated, publish-ready article. Here's the full build. In this tutorial I build a complete research-to-publish pipeline inside Moxt — an agent-native workspace where each AI teammate is a role you define once and reuse forever. No nodes, no JSON, no code. I describe the pipeline in plain English to Momo (Moxt's assistant) and it builds the workflow, the task board, and the review gates for me. The AI research team is five roles plus an illustrator: a Research Leader that turns a topic into an outline with candidate sources, a Project Planner that splits the outline into section briefs, Research Assistants that run in parallel — one per section, each pulling its own citations — an Editor on a stronger model that merges everything into one voice, an Illustrator that generates a hero image and section diagrams, and a Publisher that renders the final HTML article. We run a real topic through it live ("Why GPUs power AI, and what's changed recently"), pause at both human review gates, and watch a finished, cited, illustrated article come out the other end — then get written as a row into a Research Library you can filter before you start your next piece. Then the part most AI writing videos skip: I put the team's article side by side with a single-agent version of the same topic, same class of model, same prompt. Not to dunk on one model — to show exactly where structure beats effort: verifiable citations, real depth per section, and an article that's actually ready to ship instead of a first draft you still have to format. Finally, the second run — where the teammates already have context, the library has history, and nothing starts from blank. That compounding is the whole reason to build an AI research team instead of prompting harder. If you've seen my n8n research agents video, this is the same five-agent pipeline built a completely different way. Both links below. TIMESTAMPS: 0:00 The finished AI-written article (cited + illustrated) 2:13 Building the 5 AI teammates (Step 1) 4:09 Describing the workflow to Momo — no nodes, no code (Step 2) 5:27 Running the pipeline: outline → parallel research → edit, illustrate & publish (Steps 3–5) 9:40 Single AI agent vs. the 5-agent team (side-by-side) 12:39 The second run: why the work compounds 15:17 Running the whole team from Slack + how to fork this workflow Resources: Try Moxt: https://moxt.ai/ Moxt docs / what's new: https://moxt.ai/whats-new My n8n research agents build: https://www.youtube.com/watch?v=LSPmrCX_cJI 📧 FREE Weekly AI Automation Newsletter: Get weekly tutorials, free workflow templates, and the latest AI news straight to your inbox 👉 https://aifornoncoders.com/newsletter 🤝 Join the AI Automation Engineering Community: Go deeper — interact with me and the community, get feedback on your builds, and stay ahead of the curve 👉 https://www.skool.com/ai-automation-engineering-3014 Join this channel to get access to perks: https://www.youtube.com/channel/UCINkul2kgQYqt4GuD-352DQ/join If you build a version of this for your own content, post it in the community — I'll take a look and tell you where I'd tighten it. And if you're deciding between wiring this in n8n or describing it in Moxt, watch both videos before you pick; they solve the same problem very differently. #Moxt #AIAgents #AIResearchAgents #ContentAutomation #n8n Join this channel to get access to perks: https://www.youtube.com/channel/UCINkul2kgQYqt4GuD-352DQ/join