Laras AI — AI-Powered Decision Support for Public Services
An AI-powered decision support system built for government, turning social welfare data into precise recommendations and real actions — implemented on Manunggar Raharja, Yogyakarta's integrated poverty alleviation platform.
Demo
Overview
Laras AI is an artificial intelligence system designed to help government work faster, more accurately, and data-driven in public service decision-making.
The core premise is simple: data should not stop as reports, dashboards, or meeting materials. It should produce decisions and real actions.
Laras AI was first implemented on Manunggar Raharja — the integrated data platform for poverty alleviation in the Special Region of Yogyakarta (DIY). Through this system, Laras AI processes social assistance data comprehensively: understanding each citizen's assistance history, detecting potential targeting errors, and providing more precise follow-up recommendations.
What the System Does
From Data to Decision
Most government systems collect and display data. Laras AI goes further by turning that data into actionable intelligence:
- reads patterns across large volumes of social welfare records
- surfaces insights that would take analysts hours to find manually
- generates structured recommendations for field operators and decision-makers
- connects data, policy rules, workflows, and community needs in one orchestration layer
This shifts the role of data from passive record-keeping to active decision support.
Social Assistance Data Processing
On the Manunggar Raharja platform, Laras AI handles:
- comprehensive processing of social assistance recipient data
- reconstruction of each citizen's full assistance history across programs
- cross-referencing data to detect potential targeting errors (wrong recipients, duplicates, missed beneficiaries)
- generating precise follow-up recommendations based on verified data
The goal is to reduce errors in aid distribution and ensure that assistance reaches those who genuinely need it.
Anomaly Detection & Investigation
Laras AI includes proactive anomaly detection capabilities:
- flags statistically unusual patterns in the data
- surfaces potential data deviations for further review
- provides context around anomalies to support human investigation
- helps reduce both under-coverage and over-coverage in assistance programs
Natural Language Interface
Operators can interact with the system through a conversational chatbot interface:
- ask questions about data in natural language
- get structured answers without needing to navigate complex dashboards
- understand program context and citizen profiles quickly
- reduces the knowledge barrier for non-technical government staff
Core Features
Laras Analis
Reads patterns and trends across datasets. Surfaces insights about program coverage, beneficiary distribution, and historical changes over time.
Laras Decision Support
Provides structured recommendations for decisions — from eligibility assessments to intervention prioritization. Bridges the gap between raw data and actionable policy.
Laras Investigator
Detects anomalies and potential data deviations. Supports audit workflows and helps identify cases that require human follow-up.
Laras Chatbot
A conversational interface for operators. Answers questions naturally, explains data context, and helps non-technical users navigate complex information.
Supporting Features
Beyond the four core modules, Laras AI also includes:
- Visual dashboard — summarizes key metrics across the assistance ecosystem
- Data statistics — aggregated views for management and reporting
- AI-powered beneficiary profiles — enriched citizen profiles generated from multi-source data
- Community report management — handles incoming public reports and flags for review
Key Outcomes
- A unified AI layer on top of existing government data infrastructure
- Reduced manual effort in identifying targeting errors and anomalies
- Faster, more consistent decision-making for field operators
- A replicable model for AI-assisted public service orchestration
Screenshots
Tech Stack (At a Glance)
- AI/ML pipelines for pattern recognition and anomaly detection
- Natural language processing for the chatbot interface
- Integration with Manunggar Raharja data infrastructure
- Role-based access for operators and administrators
- Government-compliant data handling
(Technical specifics are intentionally kept high-level.)
Notes
This project is showcased as a government AI system.
Sensitive data, internal configurations, and citizen information are not disclosed.
The focus is on system architecture, decision support logic, and public service impact — not on exposing implementation details.
One-Line Summary
A government AI system that transforms social welfare data into precise decisions — and decisions into real action for the community.