一位开发人员正在进行一个企业研发项目，专注于为顾问（特别是开发角色）启用AI。该项目涉及容器化、用于镜像创建的CI/CD自动化以及构建文档。工作的关键方面是确保AI代理是沙盒化的，并与各种LLM兼容，跨不同系列进行测试以防止用户意外。 AI
影响 关注AI代理的实际实现和测试，强调LLM兼容性方面的挑战以及对强大沙盒化的需求。
排序理由 开发人员讨论为AI项目构建基础设施和功能，包括容器化和代理沙盒化。
在 Mastodon — mastodon.social 阅读 →
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
At any rate, got the initial containerization and (GitLab #CI based) image-creation/publishing automation built. Even got the first-pass "documentation" built. Now I gotta go back and do cleanup tasks. #AI
2) Even the ones that work with multiple families, generally want to test them with each of the families they support to better-ensure "no surprises" when handed over to their users. #AI
1) Some #agents are (currently?) more-limited than others with which #LLMs they'll deign to work with #AI
There's three of us doing the technical work on the project. I got the "make the agents sandboxable" portion of the workload. It's amazing how quickly one can burn tokens while writing containerization and associated #CI workflows. #AI
Recently been working a "corporate R&D" project. Company wants to #AI enable the workforce, particularly those consultants who are doing "developer" types of tasks.