I Built an AI Research Team that Self-Verifies Every Claim

Learn how to build a multi-model AI deep research workflow with Hermes Agent's new Bot Mode — three independent AI agents research the same question with zero shared context, then a fusion agent synthesizes every claim and Supabase persists it as an auditable database record. In this tutorial I build the whole thing live: a workflow architect bot that designs the agent team, three research scouts on DeepSeek, GLM, and MiniMax, a Supabase agent for persistence, and a real test on a Kitchener-Waterloo real estate investment question. 👉 Get started with Supabase for free: https://supabase.plug.dev/KQuvciP Here's what you'll see in this Hermes tutorial: I start with the finished result — a full deep research run on KWC real estate with confidence scores, 91 verified claims, and 74 citations. Then I break down exactly how the multi-agent workflow was built: creating a Workflow Architect bot (Art) that turns a project spec into a deployable skill, building the Supabase agent (Sue) with its skill install and MCP OAuth connection, and running the Deep Research Diamond — three isolated scouts researching independently so agreement across three different models means real verification, not a coin flip. The fusion agent buckets agreements and disagreements without averaging, and every source lands in Supabase so you can audit any claim and track trends across runs over time. This workflow works for any research question where accuracy matters — market analysis, competitive research, due diligence — not just real estate. TIMESTAMPS: Build a multi-agent Hermes Agent workflow that performs deep research with 91 verified claims and zero shared context. This setup uses a Workflow Architect to design the team, deploying three independent scouts running on different models to ensure verification. By isolating each agent, we get agreement across models rather than a coin flip, allowing a fusion agent to synthesize results into a final, auditable report. Everything lands in Supabase, creating a permanent database record for every claim. Whether you are handling market analysis or due diligence, this robust AI research workflow provides verifiable output. I break down the entire multi-agent system live, from the initial bot design to the persistence layer, showing you how to scale this pattern for any task where accuracy matters. Subscribe for weekly AI automation breakdowns, and comment which research use case you want to see automated next. 0:00 - The 3-AI Research Result First (Hermes Bot Mode) 1:06 - Kicking Off the Workflow Architect Bot 1:58 - How the Architect Builds the Agent Team 2:29 - Inside the Deep Research Diamond Skill 3:56 - Running the Research Workflow Live 5:51 - The Results: 91 Claims, 74 Citations, Full Verification 7:59 - What Is a Hermes Bot? (Agents Explained) 8:17 - Building the Workflow Architect Bot (Art) 9:49 - Building the Supabase Agent (Sue) + MCP Setup 10:57 - Recap: The Full Deep Research Workflow 🎁 To download this research template, join free AI Automation Hub: 👉 https://www.skool.com/ai-automation-community-5707/free-skill-hermes-deep-research-diamond 🎓 Hermes Deep Dive Course (Udemy): Master building autonomous AI workflows with Hermes Agent 👉 https://www.udemy.com/course/hermes-agent-masterclass-build-autonomous-ai-workflows/?couponCode=HERMES_AGENT Resources: Hermes Agent: https://hermes-agent.nousresearch.com/docs Supabase: https://supabase.com 📧 FREE Weekly AI Automation Newsletter: Get weekly tutorials, free workflow templates, and the latest AI news straight to your inbox 👉 https://aifornoncoders.com/newsletter #HermesAgent #AIAgents #DeepResearch #Supabase #AIAutomation