Company reportedLive since May 2025Added Oct 2026

Taiping Subdistrict, Pulandian District (Dalian)

Dalian's Taiping Subdistrict runs DeepSeek on local servers and reports office efficiency up more than 40%

A neighbourhood government office in Dalian, China, installed DeepSeek on its own servers inside a closed internal network. A news report based on the district publicity department says office efficiency rose by more than 40% and response to citizen requests by about 30%.

Key figures

Copied exactly from the sources. Copy any figure with its citation.

Up more than 40%

Office efficiency after introducing the local DeepSeek government system1

Up about 30%

Response speed on citizen requests, where AI sorts requests from several channels1

Can reduce manual checking time by 70%

Effect of automatic verification on manual data-checking time1

What it took

CostNot disclosed
TeamNot disclosed
TimelineNot disclosed

Cost, team size and timeline were not disclosed in the sources.

The problem

Taiping Subdistrict is a neighbourhood-level government office in Pulandian District, Dalian, in northeast China. Its staff draft documents, analyse policy, handle requests from residents and check data. The office wanted to raise the efficiency of this routine work while keeping sensitive information under its own control.

What they built

The office worked with Shanghai POP, an outside provider, on a local installation of DeepSeek, an openly released language model. It runs on local physical servers and a closed internal network, which the report says keeps important information from leaving the area or being passed on. Staff use it for drafting documents, summarising policy, sorting resident requests and checking data.

How they did it

Shanghai POP tailored the setup for the subdistrict. It also sent AI product designers from its head office to train the office's staff, with a theory session and a hands-on session. The training covered the system's basics, its main strengths, real government uses and tips for daily operation.

Results

According to the report, office efficiency rose by more than 40%. Response to resident requests became about 30% faster because the system gathers data from several channels and sorts the requests. Automatic verification can reduce the time spent on manual checks by 70%. The office plans to explore further uses, including local industry analysis and agriculture.

Lessons for your project

  1. Choose local hosting when data is sensitive. The office kept information inside its own servers and closed network so it does not leave the site.

  2. Pair the install with training. The vendor taught staff the basics, the main strengths and real government uses before daily use.

  3. Start with routine tasks. The reported gains come from drafting documents, sorting requests and checking data.

  4. Ask for the baseline. The report gives no starting figures, so measure your own before and after.

Caveats

  • The figures come from a news report based on information from the district's publicity department, so they are the local authority's claims. No method or baseline is given.
  • The article does not say which DeepSeek model or version is used, or how many staff use it.
  • The article is dated 2025-05-09 and describes the partnership and rollout without a go-live date. The deployed date shown is the article date.
  • The report is written as an announcement and does not describe any problems.

Sources

Summarized in our own words from the sources above. Spotted an error? See corrections.