Anyone else having issues with @OpenAI Dot working reliably right now? Sometimes the phone works and I'll message my Dot, and sometimes it never responds.

A vault answers what the secret is. Authorization answers whether this caller can use it on this run. Mixing those two up is how agents cross tenant boundaries without ever seeing the raw key.

If you schedule an agent, give it a per-run budget before you give it a cron. Recurring work without a ceiling turns into a recurring bill.

A correct final answer can hide repeated tool calls and failed retries. When evaluating an agent, check how it got there, how much it cost, and whether it knew when to stop.

An agent can remember a customer ID from last week. That does not authorize today's run to open that tenant. Memory is context. Permission is a separate decision at the tool boundary.

Has anyone run GPT Astra with a Hermes agent yet? Curious what’s working in practice and where it fell over first.

Knowing what your agent was supposed to do isn’t enough. You need to see which tools it called, what they returned, and what happened next. If you can’t reconstruct the run, how do you investigate a failure?

Open weights let attackers test prompt injections privately, outside a hosted provider’s monitoring. Your agent’s permissions need to hold even when the model follows a malicious instruction.

People get offboarded. Agents often don’t. The token still works. Role still wide. The project that created it is gone. If it can act without a living owner, it isn’t automation. It’s orphaned access.

Last week, I had the privilege of speaking at Mind and Machine DFW's first Fort Worth, TX event at @HopFusionAle . I spoke about Cybersecurity Guardrails with AI. If you're in Ft Worth and using AI in your business, be sure to come to the next event in October. mindandmachine.ai/

Looks like Grok 4.7 is coming soon... @spacexai let Grok @bot use a lower model while you figure this out! My agent team isn't working now.

The best SMB automations I see aren't "AI replaces HR." They're Sheets + Slack + a daily check, with a human still owning the weird cases. Glue first. Judgment last. Say when you’re done.

Agents don’t fail because the prompt was short. They fail because nobody defined “done.” Who can approve, and what happens when the tool lies? Spec that first. Then automate.

Reckless AI isn't mostly a model problem. It's a permissions problem. If it can email, pay, or access customer data, treat it like a new hire with root access.

Doodlenote.ai is our latest free (open-source) tool, and @AndrewWarner reviewed it in a GitHub repo review yesterday! Check out his video and let us know what you think of Doodle Note! youtu.be/rVoAyD2Zrfs?si=g-Rf…

What's going on with frontier models today? Time to invest in some hardware to run local AI Models.

We opened the first Onyx Dev Labs repo. DoodleNote. It captures the call from your computer. No bot. Transcribes on-device. Turns that plus your rough notes into something you can actually use. Audio stays put. github.com/Onyx-Dev-Labs/doo…