• Conversation: LLMs and the what/how loop A conversation between Unmesh, Rebecca, and Martin on how LLMs help us shape the abstractions in our software. • We view our challenge as building systems that survive change, requiring us to manage our cognitive load. • We can do this by mapping the â whatâ of we want our software to do into the â howâ of programming languages. • This â whatâ and â howâ are built up in a feedback loop. • TDD helps us operationalize that loop, and LLMs allow us to explore that loop in an informal and more fluid manner. • 21 January 2026 Unmesh is a Distinguished Engineer at Thoughtworks, based in Pune, India.
Article Summaries:
- Summary
A recent discussion titled “LLMs and the what/how loop” highlights a common misconception in software development: that programming is simply a linear translation of requirements into code. The conversation argues that this view misleads both newcomers and non‑coding stakeholders, leading them to over‑emphasize perfect specifications or language syntax. In the context of large language models (LLMs), the term “human‑in‑the‑loop” is often used to suggest that the LLM does most of the coding, with humans only cleaning up errors. The speakers counter that the real challenge lies in building systems that remain robust and adaptable as requirements evolve, not merely in generating code.
Sources: