How We Improved Agentic Search
We studied real coding-agent traces, benchmarked how agents use search, and built pgr to improve what agents see first and what they do next.
Read the full articleYou might also wanna read
AI Coding Agents Learn More Than Just Code
A new framework for AI coding agents promises lasting improvement by embedding human feedback as behavioral rules, bypassing traditional mod
How Coding Agents Think, Act, and Fix Real Software Systems
AI Engineer Interview Preparation Continue reading on Towards AI »
Search-Augmented Agents Cut Token Usage by 36% and Outperform Raw File Processing
Same model, same questions. Search cut tokens 36% and won 75% of blind comparisons.
.png)
lighton.ai·21d ago
The Cheapest Tool for AI Coding Agents
A study examines coding agents' tool surfaces, revealing that cost efficiency hinges on task regime and agent design, not tool variety.
Beyond AI Vibes: Deterministic Foundations for Agentic Coding
AI coding adoption is high and trust is dropping. A testing pyramid for agents, plus reproducible production context that grounds AI in real
AgentLens: Redefining How We Measure Code Agents
AgentLens offers a holistic benchmark for code agents, focusing on user experience and trajectory. It goes beyond pass/fail metrics.
Comments
Sign in to join the conversation.
No comments yet. Be the first.