Software Engineering

Explaining errors and drafting first passes

AI is used to explain errors, sketch alternatives, find forgotten APIs, and produce a first version of code or migration scripts so developers can move faster from idea to reviewable draft.

Why the human is still essential here

Humans must judge whether the explanation and draft are right, adapt them to local constraints, and ensure tests, rationale, and observability match production needs.

How people use this

Stack trace explanation

AI reads compiler messages, runtime exceptions, or failing test output and translates them into likely root causes and concrete next debugging steps.

ChatGPT / Claude

Migration script first draft

AI produces an initial SQL, framework upgrade, or API migration draft that developers can adapt for their schema, dependencies, and rollout constraints.

Claude / GitHub Copilot

Library usage lookup

AI suggests relevant methods, parameters, and example snippets for unfamiliar frameworks or SDKs so engineers can get to a reviewable first implementation faster.

ChatGPT / GitHub Copilot

Need Help Implementing AI in Your Organization?

I help companies navigate AI adoption -- from strategy to production. Whether you are building your first LLM-powered feature or scaling an agentic system, I can help you get it right.

LLM Orchestration

Design and build LLM-powered products and agentic systems

AI Strategy

Go from idea to production with a clear implementation roadmap

Compliance & Safety

Build AI with human-in-the-loop in regulated environments

Related Prompts (4)

Latest community stories (1)

Opinion
Blog

AI productivity gains are real, but smaller in production

A new meta-analysis on GenAI coding assistants landed on arXiv this month, and I think it is a useful cold shower for both sides of the argument.

The paper looked across 23 studies and found a statistically significant productivity gain from GenAI assistance in programming. Not magic. Not fake. A real effect.


But the effect was moderate, highly context-dependent, and smaller in open-source and enterprise settings than in controlled experiments. It also found no statistically significant learning gain.


That is basically the whole AI coding debate in one sentence: the tools help, but the demos are not the work.

PV
Paulo Victor Leite Lima GomesSr Eng Manager at Nubank
May 24, 2026