evidoria

← Back to browse

Good practice Imported

VNPT's AI Document Pipeline — Ho Chi Minh City's Computer-Vision and RPA System for Administrative Filings

Vietnam · Ho Chi Minh City · See the Vietnam profile · See the Ho Chi Minh City profile

Evidence: Descriptive / self-reported Top 92% 27/100 · Ask Evidence Copilot about this practice

VNPT's AI pipeline for Ho Chi Minh City's ward offices used computer vision and RPA to screen and route citizen filings; state media reported 10,000-13,000 daily paper submissions saving over 1,400 staff-hours a day, but no independent audit or error-rate data has been published.

10,000-13,000
Daily in-person filings processed (May 2025)
5,000-6,000
Daily online filings processed (May 2025)
1,400+
Staff-hours saved per day (in-person) (May 2025)
550+
Staff-hours saved per day (online) (May 2025)
~175
Full-time-staff equivalent of hours saved (May 2025)

Details

Maturity
Scaling
Promoter
Vietnam Posts and Telecommunications Group (VNPT), with Ho Chi Minh City People's Committee
Period
Piloted from late 2024; scaled from May 2025, coinciding with Vietnam's July 2025 ward-level administrative reorganisation
Keywords
administrative processing, computer vision, RPA, e-government

Context

In 2024, Ho Chi Minh City's government worked with VNPT (Vietnam Posts and Telecommunications Group) to deploy an AI-based administrative document-screening system in ward-level offices, using computer vision and RPA to screen and route citizen filings.

Activities

The pilot began in late 2024 and was scaled up from May 2025, coinciding with Vietnam's July 2025 administrative reorganisation that merged ward-level administrative units nationwide.

Results

Vietnamese state media reported the system processed 10,000-13,000 in-person filings and 5,000-6,000 online filings per day, saving over 1,400 (in-person) and 550 (online) staff-hours a day — equivalent to roughly 175 full-time staff.

Conclusions

Both reporting outlets are state media citing the same underlying city and VNPT figures from a single reporting period (May 2025); no independently audited accuracy rate, error rate, or citizen-satisfaction survey has been published, and Hanoi's parallel plan to give 'every civil servant an AI assistant' by 2026 remains a stated goal, not a demonstrated replication.

Implementation
Indicative cost
Medium (€50k–€500k) — No public cost or budget figure disclosed in available sources.
Time to results
Short (< 1 year) — Piloted from late 2024; scaled from May 2025, coinciding with the July 2025 nationwide ward-level administrative reorganisation.
Staffing & skills
VNPT technical team, Ho Chi Minh City ward-level office staff

Conditions for success

  • Integration with ward-level administrative offices and existing filing intake channels
  • Combined computer vision + NLP for document screening with RPA for intake automation
  • Scale-up timed to coincide with Vietnam's July 2025 ward-level administrative reorganisation

Common failure modes

  • No independently audited accuracy, error rate, or citizen-satisfaction data published
  • Reporting comes from two state-media outlets citing the same underlying city/VNPT figures from a single period, limiting independent verification

Where it fits

Governance type
national telecom SOE (VNPT) partnered with city government
Scale
Ho Chi Minh City ward-level offices, 15,000+ filings/day
Income level
lower-middle-income (Vietnam)

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful