Case Study

Custom GPTs for Organizational Efficiency

Viewing Case Study 01

Role / Organization

Role: Editorial & Digital Strategy
Organization: Harvard Magazine

I designed and deployed a suite of custom GPT tools to automate recurring editorial, accessibility, research, SEO, and production workflows—reducing time spent on remedial tasks while improving consistency and turnaround speed.

70–80% reduction in recurring production time
10+ Custom Built GPT Solutions
3–8 hours saved per week (volume-dependent)

Overview

The goal was not to replace editorial judgment or content creation, but to eliminate repetitive, time-intensive steps that pull attention away from higher-value work.

A secondary goal was to help educate staff on SEO initiatives, alt text/accessibility standards, and consistent organizational practices.

Why it mattered

Repetitive tasks eat away at a small staff's bandwidth across a fast-paced newsroom. The approach focused on discovery, building and testing custom solutions, and educating colleagues on how to incorporate these practices into their workflows. 

The problem: repetitive, remedial tasks

Digital media workflows often involve:

  • Repetitive remedial tasks
  • Manual accessibility work (alt text, captions)
  • Time-consuming archival research
  • SEO strategies
  • Image processing and file standardization

Individually these tasks seem small—but collectively they consume hours each week and introduce friction, inconsistency, and context switching.

The solution: a suite of custom GPTs

I built 10+ custom GPTs, each designed around a specific repeatable workflow and aligned to Harvard Magazine’s editorial standards and accessibility guidelines.

  • Automated alt-text generation aligned with accessibility requirements
  • Video caption cleanup + transcription formatting
  • Email newsletter archival article discovery + thematic curation
  • News headline aggregation for weekly editorial use
  • HTML newsletter formatting
  • SEO title, URL, and metadata rewrite suggestions + scoring
  • Image-to-text conversion with formatting (bold/italics/underlines)
  • Web-ready image conversion with publication-specific naming conventions

Each GPT functions as a reusable tool embedded directly into daily editorial work.

Results & impact

Conservative estimates based on standard workflows:

  • Average time savings: 80–85% per task
  • 15–40 minutes → 2–5 minutes for tasks that were previously manual
  • Estimated 3–8 hours saved per week (depending on production volume)

Beyond time savings:

  • Greater consistency in accessibility and formatting
  • Faster turnaround without quality tradeoffs
  • Reduced context switching and manual error
  • Scalable infrastructure that improves as volume increases
Summary

This project demonstrates how AI can benefit organizational infrastructure, not just deliver one-off productivity wins. By identifying repetitive remedial tasks, and designing GPTs around them, I was able to significantly reduce operational bottlenecks.

These tools and methods can be adapted to any digital content-heavy organization prioritizing quality, accessibility, and speed.