Smart Nation Citizen Life-Milestone Grant Routing Agent with Google ADK
Build an interactive multi-agency grant evaluation agent using Google ADK and Gemini 2.0 to navigate LifeSG family milestone services. Automatically determine eligibility across housing, childcare, and healthcare support schemes.
What you are building
The core problem, expected build, and operating context for this challenge.
Deploy Google ADK with Gemini multimodal capabilities to construct a LifeSG milestone grant evaluation and workflow routing agent.
How work is evaluated
Evaluates grant eligibility calculations and recommended action cards for citizen queries.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
correct_scheme_matched
Verifies that eligible newborn schemes are correctly identified
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
recall
Recall of applicable government support programs • target: 1 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Understand Google Agent Development Kit (ADK) core abstractions and agent loops
Leverage Gemini 2.0 models for structured intent classification over multi-agency policy documents
Implement multi-criteria evaluation algorithms matching family household profiles to Singapore grant criteria
Construct friendly, accessible citizen guidance outputs adhering to GovTech design standards
Reference links and supporting material
Synthetic dataset of household demographic profiles and public grant criteria.
Challenge setup
Evaluates grant eligibility calculations and recommended action cards for citizen queries.
Challenge input
JSON object with household_profile and citizen_query
Google ADK
Provides official Google framework tools for Gemini agent workflows.
Evaluated output
JSON object with eligible_schemes array and estimated_total_support_sgd
- Verifies that eligible newborn schemes are correctly identified
- Recall of applicable government support programs • target: 1 • range: 0-1
- Benchmark: Public Service QA Benchmark
- Recall target: 1
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
- Google ADK · Required
- Google · Optional
- Google Gemini · Optional
Evaluation contract
- correct_scheme_matched · Weight 1
- recall · Weight 1
Metrics
- recall · Target 1
Benchmarks
- Public Service QA Benchmark
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON object with household_profile and citizen_query
Output: JSON object with eligible_schemes array and estimated_total_support_sgd
Execution harness: Python
Evaluation tools
- Google ADK · Agent Development Framework
Recipe freeze marker
No recipe freeze time is recorded.