Qianlin Intelligent Brain · Final Demo Recording Script (Full Version)

Based on full E2E verification capabilities + ready demo data, total duration approx. 2 minutes. All three scenario AI dialogues use L1 deterministic routing (auto_execute=true), not relying on LLM inference, reproducible stably.

1. Pre-Recording Checklist (Mandatory)

2. Three Verified Scenarios for Reference (L1 Deterministic Routing)

| Scenario | AI Dialogue Input | L1 Rule | Tool | Expected Output |

|---|---|---|---|---|

| 1. BOM Impact Analysis | Query the impact scope of P-1001 | bi_change_impact (p9) | analyze_change_impact | 3 BOMs (ASSY-A001/002/003) + 4 parts + Medium |

| 2. Data Quality Inspection | Check Part data quality | bi_data_inspect (p8) | inspect_quality | Five-dimensional scoring + Issue statistics |

| 3. Governance Record Traceability | View recent governance records | bi_gov_history (p8) | get_governance_history | 20 audit logs |

Supplementary UI pages (all data validated)

• Quality KPI Dashboard: 10 KPI cards + 185 alerts (System 80 + Employee 105, high-level "SCSAI request failed")
• Data Asset Management: 328 assets / Part 200 / Vendor 80 / Doc 48
• Dashboard Homepage: Part 1223 / Pending ECR 28 / 62 employees online

III. Part One: Opening · Authenticity Verification (0:00-0:15)

| Time | Screen | Voiceover/Subtitles |

|---|---|---|

| 0:00-0:03 | Black screen, white text: Qianlin Intelligent Brain · Live System Demo | Light background music starts |

| 0:03-0:08 | Browser input ylxt.chat → Enter, address bar visible throughout | Subtitle: 43 industrial digital employees, experience directly without login |

| 0:08-0:15 | Homepage digital employee matrix, showing "62 agents online" / Universal Object Creation Engineer / Process Optimization Steward / SPC Monitoring Steward | Voiceover: This is not a demo, but a live product that is online and ready to experience anytime |

Operation: Simply open the homepage; no clicks required.

IV. Part Two: BOM Change Impact Analysis (0:15-0:55)

| Time | Visual | Voiceover/Subtitles |

|---|---|---|

| 0:15-0:20 | Click left-side 'Data Overview' → Dashboard homepage: Total Parts / Pending ECRs (Real-time) | Subtitle: Currently managing a large number of product Parts and pending changes |

| 0:20-0:28 | Cut to 'AI Assistant' chat interface | Subtitle: Scenario 1 · BOM Change Impact Analysis |

| 0:28-0:38 | Keyboard input 'Query the impact scope of P-1001' and press Enter | L1 bi_change_impact direct pass-through, no LLM reasoning required |

| 0:38-0:48 | Keep waiting state (1-3s real response time); AI returns results | Subtitle: Full-link impact analysis completed in 10 seconds |

| 0:48-0:55 | Freeze frame, highlight '3 BOMs + 4 affected parts' | Voiceover: One parameter change, automatic risk simulation across all work segments... 2 days → 10 seconds |

Note: Do not cut the response waiting time; show the real duration. Subtitle numbers '3 BOMs / 4 parts' in light, high-contrast highlight.

V. Part Three: Holistic Data Asset Governance (0:55-1:35)

Dimension 1 · Data Quality Inspection (0:55-1:08)

Dimension 2: Industrial Asset Governance (1:08-1:20)

Dimension 3: Risk Warning Closed Loop (1:20-1:35)

VI. Part Four: Governance Closed Loop + Conclusion (1:35-2:00)

| Time | Screen | Voiceover/Subtitles |

|---|---|---|

| 1:35-1:42 | Back to AI Assistant, inputting "View recent governance records", L1 get_governance_history returns 20 items | All governance operations are traceable |

| 1:42-1:48 | Flashback to three scenes: BOM Impact → Quality Inspection → Governance Records | Superimposed text: 10s / 10,687 entries / 20 entries |

| 1:48-1:53 | Flash to Digital Worker Matrix (Data Inspector / Data Repairer / Change Analyst) | 43 digital workers, running on edge, data controllable |

| 1:53-2:00 | White background, black text: Qianlin ZhiNao · ylxt.chat No-login Trial | Enabling Guizhou Phosphorus Chemical Industry to use its own industrial AI agent |

Closing caption: SCSAI PLM + Local SQLite + Rule Engine, core data 100% local storage.

VII. Unified Visual Design for Editing

AI巡检 ─右滑→ 资产平台 ─上推→ 风险看板
   ↓            ↓            ↓
"质量评分"    "资产治理"    "风险监控"

数字标注规范

| Scene | Annotation | Font |

|---|---|---|

| Dashboard | Part 1223 · ECR 28 | Large, bold, fade-in |

| BOM Impact Analysis | 3 BOMs · 4 Parts | Bold subtitle sync |

| Data Quality Inspection | 10,687 rules · Five-dimensional scoring | High-contrast labels |

| Governance Records | 20 audit logs | White text on dark background |

| Digital Worker Matrix | 43 workers · 62 online | Fade-in animation |

Bottom line: All alternatives must be real system screenshots or curl outputs, no prototypes or mockups allowed.

VIII. Supplement, Create BOM + Concurrency Proof (New scene, core demo)

Prerequisite: This session has successfully tested this pipeline (chat → L1 bi_bom_create → create_object → type parsing Part → confirm → resume → SCSAI creation successful).

Measured real output: Main Part (id) + Child Part + Part BOM relationship, all three persisted to database.

| Time | Scene | Operation | Core Proof |

|---|---|---|---|

| 0:35-0:38 | AI Assistant Chat Interface | Input: Create a BOM for a yellow phosphorus refining section, press Enter | Create command → L1 hits create_object |

| 0:38-0:42 | Keep confirmation card (type Part/BOM structure) waiting | Click "✅ Confirm Creation" | Manual review confirmation, compliance not skipped |

| 0:42-0:45 | Immediately switch to "Governance Audit Log" page after confirmation | Do not wait for creation to complete | ⭐ Proof of concurrency: BOM is created in the background after confirmation |

| 0:45-0:52 | Audit log scrolls (20 entries) → Asset dashboard (328 assets) | Normal browsing | Governance closed-loop and creation run in parallel |

| 0:52-0:58 | Return to AI Assistant, input: Query the BOM just created | Query | Returns the newly created Part + BOM components |

| 0:58-1:02 | Highlight returned results: components + BOM relationships | — | BOM has already been fully formed in the background |

Editing Notes:

• The confirmation card "Type: Part" must be captured to avoid misleading the audience into thinking it's a background write without confirmation.
• Edit the three segments "Confirm → Switch away → Return to query" into a seamless flow to demonstrate parallelism.
• If a Model relationship error occurs (SCSAI FK non-fatal), it can be cut or explained with a subtitle: "Only model relationship skipped, main structure already built."

8. Supplement 2: Part Release Closed-loop (New optional scenario, review bonus points)

Prerequisite: The release chain has been successfully tested in practice——chat → L1 bi_approve_item → approve_item → promoteItem → SCSAI PE_ManualRelease (Sciot PLM identity demo) → Part changes to Released.

The previous conclusion that the 'status was controlled and could not be released' was due to a permission issue and has been overturned (see the 2026-08-02 fix record for details).

| Time | Screen | Operation | Core Proof |

|---|---|---|---|

| 0:15-0:20 | AI Assistant Chat Interface | Enter 'release PART-1780803072387' and press Enter | Release command → L1 hit approve_item |

| 0:20-0:28 | Keep real waiting (1-3s) | Wait for SCSAI to perform the actual release | Status changes from Preliminary → Released in real flow |

| 0:28-0:33 | Highlight returned Released/is_released=1 | — | Genuine success: Real verification on the SCSAI side |

Editing Note:

• If the target Part is already released, the system will prompt 'must be in Preliminary/Manual Change status'. This is normal interaction; simply switch to an unreleased target (this part can be cut).
• Execute using demo/123456 (Sciot PLM identity); root is for query only.

VIII. Risk Mitigation Plan

| Risk Scenario | Contingency Plan |

|---|---|

| SCSAI server unreachable | Use pre-saved curl result screenshot, labeled 'Pre-run result in isolated network environment' |

| Inspection timeout (large Part data volume) | Pre-run the inspection once before the demo to show cached results; or specify a smaller itemType |

| Audit log is empty | Perform a BOM impact analysis before the demo to automatically generate governance records |

| AI chat routing failure (L1 not hit) | curl:POST /api/ai-agent/execute {"toolName":"analyze_change_impact","params":{"target_id":"P-1001"}} |

Bottom line: All alternatives must be real system screenshots or curl outputs; prototypes or mockups are prohibited.

IX. Final Recording Checklist

• Total duration strictly ≤ 2.5 minutes (including the new BOM creation scenario).
• The address bar 'ylxt.chat' must be clearly visible throughout.
• Dashboard displays Part 1223 / Pending ECR 28 (real-time).
• AI chat 'Query the impact scope of P-1001' returns 3 BOMs + 4 parts.
• AI chat 'Check Part data quality' returns a five-dimensional score.
• AI chat 'View recent governance records' returns 20 audit logs.
• AI chat 'Create a BOM for the yellow phosphorus refining section' → Confirmation card (type Part) → Cut away after confirmation.
• AI chat 'Release PART-1780803072387' → Released (optional scenario, bonus for review).
• After cutting away, query 'the BOM just created' returns Part + BOM components ✅ (Tested: main item/components/Part BOM all stored in DB).
• Impact analysis results appear within 10 seconds.
• Key numbers (3 BOMs / 10,687 records / 43 employees) are clearly captioned.
• Mouse operations are natural; AI response process is complete, without frame skipping or acceleration.
• File format MP4, size ≤ 50MB, audio and video are synchronized.
• Backup version is ready.
• End with clear product entry and implementation information.