🤖 AI-Powered Reporting Assistant
The Challenge
Reviewing campaign reports manually takes time. Account managers and coordinators often have to dig through Excel files or dashboards just to answer simple questions like:
- “What’s the pacing for Campaign X?”
- “Did Campaign Y deliver in full?”
- “Can I get a summary of performance by region?”
This slows down decision-making, increases dependency on specific teammates, and limits how quickly clients get answers.
The Solution
I built an AI Reporting Agent using ChatGPT’s API, trained on our standard reporting format (CSV/Excel) and internal performance rules.
It works like this:
- Upload a pacing report.
- Ask questions in plain language: “Which campaigns are underdelivering this week?”
- The AI pulls insights instantly — from top underperformers to flight status by campaign or client.
It’s like having a 24/7 assistant that reads your reports for you and gives quick, clean answers without digging through spreadsheets.
What I Led
- Trained the AI using real campaign pacing reports and business logic
- Created a prompt library for team-friendly questions
- Built a simple UI layer using Notion and embeds for testing
- Documented use cases and shared the agent with peers for early feedback
- Iterated based on reporting edge cases (e.g., paused campaigns, missing data)
The agent is now my go-to for reporting questions — and a preview of how AI can remove the noise from campaign management.