In the news
NitiBench+: A Temporal Thai Legal Question Answering Benchmark Across 17 Domains
arXiv cs.AI · Published · 1 min read
In 30 seconds
- What happened
- Researchers released NitiBench+, a Thai legal question-answering benchmark with 1,670 items across 17 domains that tracks how laws change over time.
- Why it matters
- Matters for engineers building legal AI systems, especially those handling jurisdictions where statutes are frequently amended or repealed.
- Watch out
- Best model scores only 7.78 out of 10 on the benchmark; performance drops significantly when using current law text instead of law in force at decision date.
- retrieval
- eval
- benchmark
The patterns behind this
- Temporal Knowledge Graph Memory
- Eval-Driven Development (Agent CI)
- Error Handling and Recovery Patterns
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.