We added TOON to our MCP server, then removed it 😅
Jakub Kopecky and Jiri Spilka from Apify share what the benchmarks missed, and the five lessons we’re keeping.
Big thanks to @AgenticAIFdn for sharing it with the community 🙌
Apify tested TOON in its MCP server for 30 days, then returned to JSON.
The team found that its biggest savings came from rarely used tools, while the busiest affected tool used more tokens.
@mq3737 and @jirkaspilka share the results and lessons for MCP maintainers.
Somewhere in 150,000 daily US escort ads, there are children. Finding them is a data problem - and Canary built the tool to solve it on Apify.
Kristin Boorse (Founder, Canary) and Michelle Macias Mendez are talking about it at Run. Nov 10, SF.
Get your ticket →
Vibe coding works when your agent has the right tools.
Connect Apify MCP, build your app, ask the AI which Actors would improve it. @AiHomegirl does exactly that - and the AI surprises her every time.
MCP setup ↓
2,508 saved TikToks, and “watch later” finally has a deadline.
@fabioricotta had Grok bot catalog them and check for new saves daily. New ones go to his scriptwriting bot - Apify pulls the transcript, the bot adds frames, and out comes a script.
Saved folders, put to work 👇
Coloquei o Grok Bot pra catalogar os 2.508 vídeos que salvei no TikTok, em 43 coleções.
Todo mundo tem uma pasta de "ver depois". O problema é que o depois nunca chega.
Eu já tinha resolvido isso no X. Meus 5.294 bookmarks estão catalogados por pasta e tema num repositório no