GOAI New Intelligence Base Track · Demo Recording Script

Total Duration: 3.5 minutes | 3 Scenarios + Open Platform

Opening (0:00-0:15)

Visual: BossAgents homepage → AI Workbench

Voiceover: "BossAgents, a digital workforce orchestration platform. Trustworthy AI is not a slogan; it's an architecture. Today, we'll demonstrate three real-world scenarios—workshop composite instruction, procurement closed-loop, and inspection & repair closed-loop."

Scenario 1: Workshop Composite Instruction Collaboration (0:15-1:00)

Action: Input into AI Workbench

3号泵备件库存不足，当前仅剩2件安全库存10件，同时3号离心泵保养即将到期需要安排定期保养

旁白：

• "Based on the AgentTeams hierarchical architecture, a team of 3 Agents collaborates, orchestrated by a Team Leader."
• "With one instruction from the workshop manager, the 3 Agents automatically coordinate."
• "The first Agent—the Workshop Manager Assistant—disassembles the request into a procurement application and a maintenance reminder in 6 seconds."
• "The second Agent—the Procurement Assistant—completes price comparisons from 3 suppliers and generates a purchase order in 5 seconds."
• "The third Agent—the Equipment Administrator Assistant—generates a maintenance plan and a repair work order in 6 seconds."
• "The entire process completes in 18 seconds, with an audit log for every step."

Expected Output:

• Procurement Application: Seal SEAL-P003-08
• Price Comparison Result: Company A (Lead time 1 day / Rating Excellent)
• Maintenance Plan: 5 Task Steps
• Work Order: Assign to Technician Zhang

Scenario 2: Procurement Closed Loop (1:00-2:00)

Operation: Input in AI Workbench

采购伺服电机SM-750A 50台，预算90万，15天内到货

旁白：

• "4 Agents in Serial Closed Loop: Inquiry → Comparison → Evaluation → Approval"
• "Step 1: Inquiry Assistant generates quotes from 3 suppliers in 4 seconds"
• "Step 2: Cost Analysis completes historical baseline comparison and comprehensive scoring in 8 seconds"
• "Step 3: Supplier Evaluation completes performance verification and risk assessment in 6 seconds"
• "Step 4: ECR Review completes compliance check and approves purchase in 4 seconds—high-risk procurement decisions require manual confirmation, with humans participating as team members in critical approvals"
• "Completes a traditionally 7-15 day process in 22 seconds, with each decision step traceable"

Expected Output:

• Inquiry: 3 quotes (8200/8800/9100 RMB)
• Scoring: Comprehensive scores 92/85/78
• Evaluation: Grade A Supplier, On-time Rate 98%
• Approval: Supplier A approved, amount 410,000 RMB

Scenario 3: Inspection & Repair Closed Loop (2:00-2:45)

Operation: Input in AI Workbench

检查Part数据质量，发现并修复问题

旁白：

• "3-Agent Closed Loop: Inspection → Repair → Re-inspection"
• "Inspector completes five-dimensional scoring in 5 seconds, identifying 12 issues"
• "Repair Technician generates repair plan in 3 seconds, with 8 items auto-fixed"
• "Re-inspector verifies repair effectiveness in 4 seconds—score improves from 72 to 90, closed-loop confirmed"
• "Autonomous closed-loop, no manual driving required"

Expected output:

• Inspection: Five-dimensional scoring (Completeness 72/Accuracy 85/Timeliness 60)
• Fix: 8 auto-fixes + 4 manual
• Re-inspection: loop_status=closed, closed-loop confirmed

Audit trail display (2:45-3:00)

Action: Open in browser

https://ylxt.chat/api/agents/audit/{trace_id}

旁白：

• "Every collaboration based on the AgentTeams architecture has a complete audit trail"
• "trace_id runs throughout, each span records who made what decision and when"
• "We don't promise AI is always correct, but we guarantee every decision is traceable, auditable, and correctable"

Open platform demo (3:00-3:30, optional)

Action: Open in browser

https://ylxt.chat/api/open/docs

旁白：

• "BossAgents is not just a product, but an open platform"
• "27 API endpoints, 71 digital staff members are individually callable, each with a 'Try it' button"
• "API Key authentication + rate limiting + auditing, dual-track system—zero modification for internal frontend, external enterprises can use with their Key"
• "Three-layer decoupling: Industry × Position × Staff—the same set of 71 staff members can switch knowledge context by industry and collaboration mode by position"
• "All 15 positions have a human supervisor, human-in-the-loop is not just a slogan—execution is automatically paused at key nodes, resumes only after human sign-off"
• "MTCLAW Ascend NPU local inference—LLM stays within the enterprise network, meeting manufacturing data security requirements"

Fallback plan

Checklist

• [ ] Server is up (https://ylxt.chat/api/health returns 200)
• [ ] 3 position APIs accessible (/api/goai-demo/positions)
• [ ] LLM service available (sensenova endpoint)
• [ ] Audit log table has trace_id column
• [ ] Agent interface available (/api/agents/dispatch)
• [ ] Open platform docs accessible (/api/open/docs)
• [ ] 71 staff members queryable (/api/digital-staff/list)