Advancing Developing Growing Agentic AI Together.

6.24.26

🌶️ OPENAI TAKES INFERENCE INTO ITS OWN HANDS FOR SPEEDIER AGENTS: Engineering samples of Jalapeño are already running GPT‑5.3‑Codex‑Spark at target frequency and power after a nine-month tape-out. Initial deployment is planned by the end of 2026, with early tests showing substantially better performance per watt. The model race is becoming a silicon race, with loop engineering powering the design process.  OpenAI

🌐 DNS WANTS TO GIVE AGENTS A PASSPORT: The Linux Foundation’s proposed Agent Name Service anchors portable identity, verification, and discovery to DNS instead of a proprietary registry. Big name DNS providers are behind the project. The idea is that in the era of agents, just knowing the address is no longer enough. Traffic over the same internet but many more (and very different) users.  The New Stack

🧠 ENGRAM RAISES $98M TO BUILD AI THAT REMEMBERS WITH CONTEXT MAXING: Its 13-person team is developing compact, reusable memory trained on each organization’s documents, workflows, and institutional knowledge, claiming frontier-level results with 1–10% of the tokens. Microsoft, Notion, and Harvey are early partners. Good agents remember how you work. As Satya has said, that’s your moat.  — Engram Lab

Subscribe to the Daily Agentic AI LinkedIn Newsletter →