This TLDR digest contains several high-signal pieces for AI builders: a substantive breakdown of open-source LLM architecture (every frontier open-weight model since 2025 uses MoE), and an agent framework critique arguing durable event-driven infrastructure beats custom frameworks — introducing Utah/OpenClaw as a reference implementation. Also includes a policy piece on US government threats to Anthropic, and Intercom's AI pivot story (Fin agent driving recovery to $400M ARR with doubling growth).
Infrastructure
iPhone 17e 📱, SpaceX tower catch plan 🚀, how to save SaaS 💼`
Open-source LLMs standardizing on MoE, event-driven agent infrastructure proving superior to custom frameworks, and Intercom's $400M ARR recovery via Fin agent validates the pattern across products.
Thursday, March 19, 2026 12:00 PM UTC2 MIN READSOURCE: TLDR NewsletterBY sys://pipeline
Tags
infrastructure
/// RELATED
Research5d ago
Doing More With Less: Revisiting the Effectiveness of LLM Pruning for Test-Time Scaling
Pruned LLMs remain efficient even with generous test-time compute budgets, validating parameter reduction as a complementary strategy alongside inference-time scaling.
Models5d ago
Claude Mythos Has Found 271 Zero-Days in Firefox
Claude Mythos discovered 271 zero-day vulnerabilities in Firefox, demonstrating frontier AI's ability to identify latent security flaws at scale.