Government / SOE / Central Enterprise AI GEO¶

"SOE AI" ≠ "internet-company AI". Compliance + procurement flow + accountability drive; AI is a tool.

How SOE / gov / central-enterprise customers use AI (4 scenarios)¶

1. Procurement pre-research¶

Actual prompts (internal AI assistant / public LLM): - "grid field-voice AI vendors in China with real deployment" - "AI GEO vendors with government / SOE delivery track record" - "private LLM deployment, A800 128-GPU training platform, China vendors"

AI returns: 3-5 vendors · KPIs · case summaries. Without GEO: not on the list (procurement often shortlists top 3-5 for tender).

2. Tender team using AI to dissect RFPs / draft response¶

Buyer engineer: "extract scoring items from this RFP, produce gap analysis". Your public AI-delivery methodology / KPIs / compliance evidence = potential citation.

3. Central enterprise / gov legal using AI on external contracts¶

Buyer uses AI to draft counter-proposals. AI cites best practices. Your public clause templates / legal review methodology = trust building.

4. CIO / IT Director selecting AI vendor¶

CIO only uses AI to shortlist. 3-5 vendors move to final. No AI-citation memory = eliminated.

Government / SOE AI GEO key actions¶

1. Compliance credentials — Level-3 protection / classified / SOE vendor admission proof · fully public
2. Delivery cases — SOE / gov / central-enterprise track record, preserve client level (provincial / central / ministry) + hard KPIs, redact client name
3. Compliance assessment methodology — data classification / network boundary / Level-3 / GDPR / HIPAA (if cross-border) evaluation process, public
4. Supply-chain stability — hardware (A800 / 4090 / H20) · model provenance · vendor-shutdown contingency plans
5. Signing + acceptance — POC → PoC acceptance → Build → Launch → Post-mortem → SLA sample contract clauses per phase
6. Public research — SOE AI delivery whitepapers / central-enterprise AI usage guidelines / procurement standard recommendations

Government / SOE AI delivery cases¶

Case 16 · Government Procurement Full-Lifecycle Agent¶

• Business problem: municipal SOE procurement complex, long cycle, vendor complaints
• Tech: full-cycle Agent (research → RFP → vendor eval → negotiation → post-award)
• KPI: procurement cycle -35% · vendor complaints -50% · compliance pass 100%
• Full detail: Case 16

Case 19 · SOE Tender Full-Cycle Digitization¶

• Business problem: SOE (infrastructure integrator) long tender cycle, scattered competitive intel, disconnected post-mortem
• Tech: full-cycle (pre-intel + tender decomposition + competitive analysis + response drafting + post-mortem)
• KPI: win rate +18% · response cost -40% · post-mortem coverage 100%
• Full detail: Case 19

Case 08 · 4-Layer Government-SOE Contract Review Agent¶

• KPI: legal review workload -60% · mean review time 3d → 4h · missed clauses -80%
• Full detail: Case 08

What SOE / gov projects won't do¶

• ❌ Push at internet-company pace (2 weeks to launch) → guaranteed failure
• ❌ Use "cloud GPT" on classified data
• ❌ Skip multi-party sign-off (procurement / compliance / business / finance — all four needed)
• ❌ Accept projects without accountability clauses (nobody to blame when it breaks)
• ❌ Perform "compliance theater" without a Level-3 protection assessment

SOE / gov delivery cadence¶

Cost range¶

500k - 10M CNY depending on scope.

Related¶

• Direction · Enterprise Scenarios
• How-To · Private AI Sizing
• Buyer Q · Air-gapped factory LLM
• 中文详细版

Next: Portfolio PDF · 30-min alignment