Tony’s Blog

LLM prompt changes are hard to verify and easy to break. PromptEval approaches this by treating prompts like code — running versioned prompts against datasets with repeatable evaluations. The key difference is reproducibility: every prompt change can be compared, measured, and traced before it ships.

Planning usually stops at estimation, leaving daily execution fragmented across tools. Scopify unifies requirements, workload estimation, and resource planning in one system. The key difference is execution visibility — projects are broken into daily workloads that leaders can arrange by drag-and-drop and immediately see capacity gaps and risks.

Traditional HR systems are powerful but inflexible, slowing organizations during change. PeopleSuite experiments with rebuilding a serious HR SaaS using an AI-first interface. The key difference is abstraction — complex data models and workflows remain intact, while users interact through assistant-style actions instead of tables, forms, and task inboxes.

Most LMS platforms are built for courses, not learning. LearnStack explores a new AI-native learning stack that rethinks how skills, knowledge, and practice are delivered. The key difference is reconstruction — replacing static learning paths with dynamic, context-driven learning designed for the AI era.

MyCareerHelp began as an early AI-powered career experiment when LLM capabilities were still limited. It is now being redesigned with modern models and refined workflows. The key difference is maturity — the same vision rebuilt with stronger reasoning, deeper personalization, and more capable AI foundations.