Pinned Tweet
things are getting very bullish in crypto, and I feel sidelined.
why?
i’m currently illiquid and jobless despite being good at what i do.
if you need a:
- customer support agent
- community guy
- customer success manager
or anyone whose job is helping people, i’m your guy.
boy oh boy.
Some news: I’m joining OpenAI to lead product management for Core Products!
I have a lot of respect for OpenAI for showing the world that scaling works, for bringing AI to over a billion people, and for believing great products are core to the mission.
I’m excited to work with @thsottiaux, an incredible PM team, and everyone across OpenAI. Together, we’ll build agents that have your back and your team’s, and keep you in control, so you get more time for the people and work you love.
Lots to learn, and lots to build. Please keep your feedback coming!
my glyph card. 2,600 $GLYPH pending in the reserve 🌊
ask glyph anything: askglyph.chat/?ref=directbra…
damian retweeted
ask $GLYPH anything 🌊
early fcfs: askglyph.chat
the whole ai vs motion designers and video editors hassle on the tl is now becoming very toxic.
“ai just killed this, ai just killed that.”
it’s more like we are all praying and hoping that everyone runs out of jobs or sumn.
we should be more focused on collaborating and exploring ways these ai tools can help us ship faster, while maintaining quality at its highest.
i crashed out on my AI agent the other day.
i had asked it to fix a bug i encountered while testing the sign-up flow on my final-year project, and it did.
the problem was, it also altered some other files in the process and broke an important feature.
i was pissed and even almost smashed my PC (we've all been there).
and this wasn't even the first time i'd clearly declared my intention to the agent, only for it to drift away from the plan.
thankfully, my anger didn't last long and this has now become a thing of the past.
i now make use of @armoriqAI's ArmorCodex to ensure that my ai agent never does anything i did not ask for.
i've attached a video walkthrough of the full setup, with the technical breakdown and what I learned in the thread below.🧵 @ArmorIQForDevs
10/ what actually broke along the way (insights)
- most of the setup friction wasn't ArmorIQ's fault, it was WSL.
- browser wouldn't open for account auth (WSL has no default browser launcher, fixed with wslu + symlinking xdg-open).
- armoriq command not found after install (npm's global prefix pointed somewhere i didn't have write access to).
- PATH fixes not sticking in new terminals (my shell reads .bash_profile, not .bashrc, so the fix was sitting in the wrong file).
- Codex itself crashed on WSL with a missing linux binary, an OpenAI packaging issue, took a pinned reinstall and a couple of retries to resolve.
- one real ArmorIQ-side thing: the installer writes an old [features].codex_hooks key instead of the current [features].hooks.
this throws a deprecation warning on every launch. cosmetic, doesn't affect function, gets overwritten on every reinstall anyway so no point patching it manually.
11/ takeaway
this started because an agent already went outside scope on my project once, and it wasn't the first time.
what stood out after actually testing ArmorCodex wasn't that it makes an agent "safe."
it's that it gives you a real mechanism, declare intent, enforce policy, check before execution, audit after, instead of just trusting the agent to stay in its lane.
