Singapore's OneService chatbot lets residents report municipal issues like potholes or broken streetlights via WhatsApp/Telegram in plain language. An NLP model auto-classifies reports with 80-85% accuracy, handling roughly 20,000 pieces of feedback a month.
450
Pre-launch usability trial participants
~80%
Case-type categoriser accuracy
~85%
Case-detail recogniser accuracy
~20,000 pieces/month
Municipal feedback volume
Details
Maturity
Established
Promoter
Municipal Services Office (MSO), Ministry of National Development, with GovTech Singapore
OneService is Singapore's municipal issue-reporting chatbot, built by the Municipal Services Office (MSO, under the Ministry of National Development) with GovTech Singapore's data-science team on GovTech's VICA chatbot platform, part of the 2019 National AI Strategy.
Objectives
To let residents report problems such as broken streetlights, potholes, litter, or illegal parking via WhatsApp or Telegram in plain language (English, Chinese, Malay, Tamil) instead of navigating fixed menus in a standalone app.
Activities
An NLP/deep-learning model classifies the issue type and extracts case details such as location and description, routing reports to the correct government agency and escalating low-confidence cases to a human agent; image recognition helps identify issues like cigarette-butt litter or streetlight defects from photos. Before the July 2021 beta launch, MSO and GovTech ran a trial with 450 participants.
Results
GovTech's technical writeup reports the case-type categoriser is correct roughly 80% of the time and the case-detail recogniser about 85% accurate, figures repeated by Cities Today and OpenGov Asia. Singapore's Senior Minister of State Sim Ann stated OneService (app plus chatbot) now receives approximately 20,000 pieces of municipal feedback per month. The chatbot remains in active use and continues to be iterated on, including a newer 'Tell Us @ OneService' beta variant.
Conclusions
Most public quantitative evidence traces to a single, comparatively small 450-person pre-launch trial plus one minister's aggregate volume figure; there is no independently audited, large-scale post-launch dataset on resolution rates or citizen satisfaction, and the reported 80-85% classification accuracy means roughly one in five to one in six reports may be miscategorised.
Implementation
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
Data sources
Where this practice's information was retrieved from, and when.
Bogotá's WhatsApp-based AI assistant Chatico logged 3.6M+ conversations since 2023, handled 1.6M queries during a water-rationing crisis, and carried 147,000+ …