Sample Use Case

Customer asks for a complex AI solution proposal

Step-by-step agent collaboration

  1. Step 1 · Coordinator Agent
    Conf 96%Low
    Tools: Task router, Workflow tracker
    Next: Sales Agent
    Input: Customer RFP for AI platform
    Output: 9-task workflow across 7 agents
  2. Step 2 · Sales Agent
    Conf 86%Medium
    Tools: CRM, Customer 360, Past Proposals RAG
    Next: Pre-Sales Agent
    Input: Workflow trigger + RFP
    Output: Qualified opportunity brief, account context
  3. Step 3 · Pre-Sales Agent
    Conf 87%Medium
    Tools: Solution templates RAG, Architecture RAG
    Next: Product Agent
    Input: Qualified brief
    Output: Solution outline v0.9
  4. Step 4 · Product Agent
    Conf 91%Low
    Tools: Product catalogue RAG, Roadmap DB
    Next: Finance Agent
    Input: Solution outline
    Output: Fitment matrix (GA vs Roadmap)
  5. Step 5 · Finance Agent
    Conf 82%Medium
    Tools: Pricing calculator, Margin policy RAG
    Next: Legal Agent
    Input: Solution + effort estimate
    Output: Pricing scenario with margin analysis
  6. Step 6 · Legal Agent
    Conf 90%High
    Tools: Contract repository RAG, Clause library RAG
    Next: IT-Security Agent
    Input: Customer redlines
    Output: Risk findings + fallback wording
  7. Step 7 · IT-Security Agent
    Conf 88%Medium
    Tools: Compliance framework RAG, Risk scanner
    Next: Operations Agent
    Input: Architecture + data flows
    Output: Control set + DPA recommendations
  8. Step 8 · Operations Agent
    Conf 78%Medium
    Tools: Resource availability, Delivery risk calculator
    Next: MD / Leadership Agent
    Input: Solution outline
    Output: Phased 14-week plan, capacity confirmed
  9. Step 9 · MD / Leadership Agent
    Conf 93%Medium
    Tools: Executive dashboard, Risk heatmap, Approval workflow
    Next: Coordinator Agent
    Input: All agent outputs
    Output: Conditional approval with 3 conditions
  10. Step 10 · Coordinator Agent
    Conf 96%Low
    Tools: Notification tool, Workflow tracker
    Next: Human Reviewer
    Input: MD decision
    Output: Final response, human approval queued