Digital transformation assistant

Consultants spend weeks turning process maps into training materials. We wanted to upload PDFs, have AI understand the workflow, then generate PowerPoint decks, scripts, or guides in minutes.

GPT-4 Vision analyses uploaded PDFs and images in parallel. Extracts roles, tasks, software dependencies, and structured summaries. Chat interface for questions. Training generator uses o3-mini with medium reasoning to produce PowerPoint (with presenter notes), training scripts (with timings), audio narration, or infographic guides. All from the same process map.

Built for a specific client project. Worked well but narrow use case. We’re not actively developing it but the approach (Vision for analysis + reasoning model for generation) is a pattern we’re applying elsewhere.