AI Call Report Automation System
An AI-powered call reporting system built to convert Fireflies meeting transcripts into clean, structured executive summaries. Designed for teams handling frequent client calls, the automation transforms raw transcripts into actionable documents stored in Google Drive and automatically shared on Slack.
Our Approach
We reverse-engineered the meeting documentation workflow: how calls were recorded, how transcripts were retrieved, and how summaries were manually prepared. Using Fireflies’ GraphQL API for transcript data and Claude Sonnet for structured reasoning, we designed a robust automation with error handling and multi-step formatting logic.
Key Features
Automated Transcript Retrieval
Retrieves call metadata and transcript content via Fireflies GraphQL endpoint as soon as transcription finishes.
LLM-Powered Summary Generation
Uses Claude Sonnet with tailored prompts to extract overview, discussion context, and action items grouped by participant.
Google Docs Auto-Creation
Generates a formatted document in Drive with headings, spacing rules, and structured output.
Slack Notification Workflow
Posts a message with meeting title, date, and buttons linking to the doc and Fireflies transcript.
Results
- Eliminated manual note-taking and reduced admin time.
- 24/7 reporting pipeline with consistent, review-ready summaries.
- Improved follow-up rates by providing timely action items to stakeholders.
Advanced Reliability Controls
- Retry logic for model errors
- Backoff & break module for pacing
- Detailed logging for traceability