# Viktor Research Engineering and research deep-dives from the team building Viktor. Index of 2 posts. Each post is also available as Markdown at `/research//md`. --- ## How We Built Viktor Around Prompt Caching How Viktor's thread engine is designed around prompt caching: SDK instead of tool loading, append-only threads, in-cache compaction, and an 80% cost cut. _Published 2026-06-08 · By Toni Albert_ - Read: https://viktor.com/research/how-we-built-viktor-around-prompt-caching - Markdown: https://viktor.com/research/how-we-built-viktor-around-prompt-caching/md ## What Breaks When Your Agent Has 100,000 Tools Most AI agents demo well and fall apart in production. Here's what we learned about tool use, context windows, memory, and UX while building an AI coworker that connects to 3,000+ integrations. _Published 2026-03-03 · By Peter Albert_ - Read: https://viktor.com/research/what-breaks-when-your-agent-has-100000-tools - Markdown: https://viktor.com/research/what-breaks-when-your-agent-has-100000-tools/md