OpenAI just launched their Agents SDK which equips LLMs with an elegant set of primitives:
• Tools (function calling)
• Handoffs
• Guardrails
• Tracing
https://t.co/ZWjVGpXdlW
Some early thoughts 🧵 /11:
Wasn't joking about this one btw
You can reverse-engineer pretty much any part of Apple platform internals in seconds using Claude or Codex with Hopper MCP
One of my favourite moments since moving to SF: went to this mango event, offered to help cut mangos for a bit, and realised, 15 mins later, I was the only one left cutting
And so I continued out of a sense of moral obligation and ended up hyperfocusing at mango cutting for 2hrs
OpenAI just launched their Agents SDK which equips LLMs with an elegant set of primitives:
• Tools (function calling)
• Handoffs
• Guardrails
• Tracing
https://t.co/ZWjVGpXdlW
Some early thoughts 🧵 /11:
OpenAI just launched their Agents SDK which equips LLMs with an elegant set of primitives:
• Tools (function calling)
• Handoffs
• Guardrails
• Tracing
https://t.co/ZWjVGpXdlW
Some early thoughts 🧵 /11:
11/11. @AgentOpsAI was a 3rd party tracer suggested by OpenAI. I tried it and liked how I can track cost, latency, prompts, and completions, pinpoint which tools get activated and why. This made tracking guardrail activations and handoffs substantially easier.