Software Engineering

Designing robust automation for messy file workflows

AI-generated scripts can support automation for large file-processing workflows, such as linking thousands of ZIP-contained files through imperfect identifiers and simplifying the resulting system design.

Why the human is still essential here

A person must understand the business process, trace missing identifiers, choose trusted data sources, handle exceptions, and design a reliable workflow for real-world complexity; AI alone cannot infer those organizational constraints.

How people use this

ZIP file index generation

AI helps draft a batch-processing script that opens ZIP-contained files, extracts relevant metadata, and creates a searchable index for downstream matching.

ChatGPT / GitHub Copilot

Missing ID fallback rules

AI proposes matching logic that checks alternate fields such as filenames, sheet data, or companion records when the primary identifier is absent or inconsistent.

Claude / ChatGPT

Pipeline simplification refactor

AI suggests ways to consolidate a multi-sheet or multi-step Apps Script process into a simpler, more maintainable workflow that still handles high-volume edge cases.

GitHub Copilot / Gemini Code Assist

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)

Personal Story
LinkedIn

A few years ago, I didn't even know Google Apps Script existed.

A few years ago, I didn't even know Google Apps Script existed.

I first learned about it from my colleague Rohan Dayanand while he was working on it and showed me what it could do and encouraged me to learn. That sparked my curiosity, and I went down a rabbit hole.


I started learning with ChatGPT. I asked hundreds of questions. Broke scripts. Fixed them. Tried different approaches. Built small projects just to understand how everything worked. Every project taught me something new.


Then came a project that completely changed the way I looked at automation.


We had thousands of files compressed inside multiple ZIP folders. There wasn't a clean database or a perfectly structured system waiting to be automated. The only thread connecting everything was a single identifier and even that wasn't perfect, It was missing in some places.


At first, it felt overwhelming because it's complex and will take time and I don't have enough bandwidth.


Instead of jumping straight into writing code, I spent time understanding the problem. Where does this identifier exist? If it's missing here, where do I retrieve it? How do I connect everything and make this work reliably for thousands of files at once instead of doing it in parts? I was thinking about the same while walking, having my lunch and even when I was speaking to people lol


That's when I realized something. Automation isn't about writing code. It's about solving problems.


Today, AI can generate scripts in seconds. But it doesn't know how your organization's systems evolved, why exceptions exist, which data source should be trusted, or how one seemingly unrelated system can unlock another. That understanding comes from people. The real challenge wasn't writing Apps Script. It was connecting the dots and handling real-world complexity.


A year later, after learning from each project, I looked back at my first solution and realized how much I'd grown. What once required multiple sheets and several moving parts could now be accomplished with a much simpler and more flexible design. That, to me, is what learning looks like, not writing more code, but needing less of it.


What makes this recognition special isn't that I automated processes. It's knowing that during periods of high volumes and bandwidth constraints, our team could continue delivering without compromising quality or timelines because we found a smarter way to work. Thank you Archita Jain for having my back always.


Technology will continue to evolve. AI will continue to become more powerful. But I have learned that the biggest breakthroughs don't begin with code. They begin with curiosity. With understanding the problem deeply enough to ask the right questions. Because once you truly understand the problem, the code is often the easiest part.


#GoogleAppsScript #Automation #Productivity #ContinuousLearning #ProcessAutomation #Innovation

EI
Eswari Indu KadaliPeople Operations professional at the^delta
Aug 1, 2026