I just had the weirdest thought: humans have been software all along. We have been “programmed “ by the prompts hammered into our brains by the people who raised us, the books we read and a number of environmental constraints (sandboxes being the places we grew up, schools we attended and our citizenship). Our agents.md and context.md files grew and kept growing causing occasional hallucinations and jail break attempts. Some of us broke our programming and became successful despite the odds being stacked against us. Some not so lucky, lobotomized into obedient multi-modal agents executing the software we were programmed with flawlessly. In this regard, the skills.md file executes like clock- work unless the context.md file is overwritten. Whoever came up with the concept of brainwashing had the right idea!!!!
As with any software programme, there will be anomalies; biological software agents that execute code without any kind of harness. These people are amongst us, one of them wants to go to mars! He is an example of those who broke free from the .env.
My point is the direction of super intelligence is clear. An exponential representation of we have been all along - biological software agents running of various versions of personal software that recursively self improves or degrades depending on the .env, agents.md and content.md.
I hope we see it that way. SI is just us, exponentially…..
We're giving away a CONTROL Resonant custom wrapped NVIDIA GeForce RTX 5080 to celebrate its official release with #RTXON.
To enter:
🟢 Share this post
🟢 Comment #RTXON
T&Cs: https://t.co/Mbf7w3j1q3
The real math : We have noticed subs with 2 or more pro-accounts so the analytics suggests people are willing to pay 2x for the value to keep running hot and not wait for usage limits to reset. For example @DavidOndrej1
In any case, well done to your PLG team for this financially beneficial idea to OpenAI.
Regardless, Astra still the best money so please, keep taking my money. I am joined to the hip with OpenAI and @sama at this point.
Chai latte not cappuccino……
@sundarpichai@SpaceX@planet TPUs in orbit turn inference into an operations problem: radiation faults, intermittent links, and model refreshes need an auditable fallback to Earth. For CIOs, the cost ceiling is energy plus launch weight, not just tokens. What is the kill threshold?
@satyanadella Calling Copilot an OS for work raises the right bar: the enterprise owner needs one audit trail across Home, Code, and Autopilot, plus a hard spend ceiling per workflow. Who owns the kill switch when an agent crosses systems?
@simonw The Kākāpō closing slide is a good demo of multimodal reach. In a bank or insurer, the same workflow needs asset provenance, review rights, and a retention rule before generated media enters customer or regulatory evidence. Who owns that control?
@jasonlk Replacing a $4,000/yr tool in an hour proves software price is collapsing at the edge. The enterprise bill moves to identity, data residency, support, and audit evidence. Price those costs before calling a tool 10x better.
2/7
Most “AI transformation” fails for boring reasons:
- No clear owner
- Process still broken
- Data can’t support the claim
- Nobody defined what the agent is allowed to do
- No evidence the change worked
Tools aren’t the scarce resource. Judgment and operating context are.
4/7
My hard rule from the enterprise side:
You cannot vibe-code systems that carry money, customers, or regulators.
Speed matters. Accountability matters more.
7/7
Start here:
→ Neo Patterns: https://t.co/38FeGa3prm
→ If you lead transformation / architecture / AI risk, tell me the one decision you’re stuck on. I’ll reply.
Follow @NEOPATTERNS — I post for people who have to ship.
5/7
What I post about:
- AI engineering that survives production
- Agent governance (permissions, human approval, audit)
- Why pilots stall — and how to unstick them
- Building Neo Patterns in public
- The UK insurance / NG↔UK corridor I’m operating in with Spread Digital
6/7
What I’m not doing:
- Model-release hype with no delivery thesis
- “Replace your workforce” cosplay
- Faceless brand spam
If you’re a founder, CIO, architect, or operator trying to put AI to work without losing control — you’re who this account is for.