AI-assisted engineering communication, reporting, and context gathering

I started using the GitHub CoPilot at work this week to connect my local files on my machine to an agentic code tool like Claude Code. @GitHubCopilot is really nice. If you are working in a corporate environment and you're a @Microsoft shop, this is a great tool with access to all the main LLM models.

I know you can use the various LLM models through VS code as well, but I'm not a developer and just want the natural language interface for developing specific solutions for myself. An example is I created an API connection with JIRA for project management, and it worked seamlessly. That solution is also much easier than PowerBI because I can ask questions of the data better and adjust the solution on the fly.

Worth checking out especially if your company doesn't allow access to Claude Code or Codex.

I expected to slow down. I didn’t expect to realize how much I’d already lost.

Let me be upfront: I’m not anti-AI. I use it every day. I’ve written about it. I’ve defended it in code reviews when teammates were skeptical.

But three weeks ago, I closed Copilot, stopped opening Claude for code, and went cold turkey for two weeks.

Not because I hate the tools. Because I started noticing something that scared me more than any production incident I’d ever worked through.

I was getting faster. And dumber.

I replaced my morning standup prep with a markdown file. Instead of opening Jira, GitHub, and Confluence every morning, I now run one terminal command and get a full status picture in ~15 seconds. I used Claude Code “skills” (a markdown instructions file that becomes a slash command) to pull from Jira/GitHub/Confluence in parallel via MCP tools and the gh CLI, with different lookback windows for Mondays vs other days. The key win isn’t code generation—it’s automating context gathering (what changed overnight, what’s blocked, what needs attention) so I can walk into standup prepared.

In my previous post I listed the top 5 things I use AI for apart from writing code.
Reena Garg asked which AI tools I use for these tasks 🙏 — sharing my list👇

⭐️ Thinking partner for ideas Tool: Cursor (Plan/Ask mode) I use Plan mode to discuss a feature/design: “what am I missing?”, “what can go wrong?”, “what’s the simplest version?”, “how should I break this into steps?”, “How can we implement this?”,

⭐️ Finding edge cases Tool: Claude (Opus-4.6) I paste the context (API, flow, PR description) and ask it to list edge cases: retries, timeouts, caching, permissions, weird inputs, etc. Opus works great for any code related tasks.

⭐️ Code review before the code review Tool: CodeRabbit, GitHub Copilot, Cursor Review These tools are really great to get a review before asking for a human review. I use them on all of my PRs.

⭐️ Turning manual processes into clean steps Tool: ChatGPT / Claude This is usually a back-and-forth conversation until it becomes a clean checklist/runbook and sometimes an automation script.

⭐️ Writing clearer communication Tool: Any AI works, but I mostly use ChatGPT PR descriptions, Slack updates, incident notes — it helps turn messy thoughts into clear writing.

If you’re using any AI tools in a different way, please share — I’d love to learn and try them too 🙌
Cheers, Princi 👩🏻💻