Salesforce AI Research Unveils TEX Test-Time Scaling Paper

Salesforce AI Research posted about its paper TEX: Test-Time Scaling Testing Agents via Execution-based Cross-Validation. The work proposes execution-based cross-validation to scale testing agents at inference time. Authors listed are Srijan Bansal, Akash Gokul, Yingbo Zhou, and Semih Yavuz. The post references EMNLP2026 findings and links to salesforce.com/blog/tex-test-time-scaling. The announcement notes the rise of software engineering agents and LLMs writing code for real-world tools such as Cursor and Claude Code.