Computers are still dumb: bringing your AI magic to enterprises
summary
DeepL runs its own models on Scandinavian supercomputers and has about a thousand employees yet integrating enterprise input still dominates the work flow. Expect to support DOCX PDF XLS HTML, OCR like Tesseract OCR, Google Docs and Sheets, Shopify/WooCommerce, GitHub and Jira, handle HTML tags and alt text with rule-based extraction, offer async job IDs with polling, export to Spotify/YouTube plus lyrics/credits and ASCAP/BMI metadata, provide client libraries across languages including PHP, and deploy regionally or use FP8/FP4 model variants to trade quality for latency.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:09APIs for scaling AI startups
- 3:15Diverse input formats
- 7:30Processing HTML for AI
- 11:43AI musicals output
- 15:15Legacy and modern languages
- 17:21Latency optimization
related talks
30mBuilding an AI-Powered Organization
18mSmall AI Teams with Huge Impact
35mDeepL: A Product Culture that Creates Top-Notch AI Products
24mDocumentation and AI: How to write right now
24mMachine Intelligence gets real
51mCan AI Face Your (Potential) Customers?
33mAgentic AI – What's Next?
01h 19mLearn to Unlock Document Intelligence with Open-Source AI