Urgent.News

What's breaking now, across thousands of outlets.

AI

LLM Prompts Need API Contracts, Not Git Diffs

TL;DR — Most teams version prompts the way they version code: line diffs and PR review. That hides the real risk, because prompt text is both the interface and the implementation at once. Borrow the API design split — define a contract for what a prompt must guarantee, separately from the wording that satisfies it — and prompt changes finally become testable. Every team that ships LLM features…

Most teams version prompts in the same way they version code: through line diffs and PR reviews. However, this approach hides the true risk because prompts function as both the interface and the implementation simultaneously. The solution lies in adopting an API design split, defining a contract for what a prompt must guarantee, and separating it from the wording used to satisfy that contract. By doing so, prompt changes can then be tested, ensuring that alterations in wording do not compromise the underlying guarantees.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Tech, AI leaders expand China's 2026 rich list

China’s 2026 rich list sees a record 1,529 mega-wealthy individuals, with tech leaders like DeepSeek’s Liang Wenfeng and ByteDance’s Zhang Yiming driving growth, reflecting the country’s expanding AI…

AI Writing Assistants for Technical Content: Who Survives?

Writing technical content is the highest-leverage marketing activity for indie developers, but it is also the most time-consuming.

  • Lex excels at producing clear, accurate technical blog posts with code examples
  • Notion AI streamlines documentation and API reference creation for teams using Notion
  • Jasper delivers best conversion-optimized marketing copy for landing pages and newsletters

Using artificial intelligence to step up the fight against doping

Tools used by WADA and national bodies, the danger that would-be cheats use them too, the personal data risk from a simple online post, and lessons from the "Santorini Summit".

  • AI can enhance anti-doping efforts by processing large data sets and identifying trends.
  • AI risks include misuse by cheaters, requiring careful harnessing of technology.
  • Athletes must understand legal implications of AI searches for legal proceedings.

More from Wednesday 28 October →