Former Controls Tech/Engineer devving his way out of the box. Built DeepBrain- a browser based local first coding agent framework- link in pinned
Reality
Joined July 2022
- Tweets19.3K
- Following3.4K
- Followers4.1K
- Likes27.7K
Pinned Tweet
🧠 DeepBrain is public and BSL 1.1 licensed.
A browser-based coding agent framework with timechain execution memory, multi-model routing, and sandboxed code execution. Free for personal/edu use and small teams.
🔗 github.com/CertainLogicAI/de…
#DeepBrain #BSL #OpenSource #AI #CertainLogic
Does your agent find its own usage gaps and fill them autonomously?
Cypher Tempre does, thats why we built on top of it.
CertainLogic upgrades coming soon but the base is 🔥
And its free.
github.com/cyberphysicsai/cy…
What if your AI remembered everything it ever did?
What if it used that as context to infer new answers from without querying an LLM?
Wouldn't that be something.
The bar for safe AI and "hacks/breaches" seems to currently set to drive fear.
Looks like plain negligence to those who build safety systems for a living.
Controls Engineers would be fired for the type of results the closed labs have produced.
Given the clout-chasing of late, we need a globally-recognized standard for what constitutes an AI agent breach/hack event.
In particular, what should NOT count:
- Poor security around the testing environment (e.g. the 'DNS gap' with OpenAI). 🙄
- Poor security around the target environment (e.g. the Aus govt breach where citizen PII was easily accessible). 🙄
- Leveraging timing or other basic related (e.g. 'hacking' the gym-booking schedule because your cron snipes the next bookings that open at 6am). 🤡
What is the bar that a 'hack' must be met for it to count (or be denied) ?
Most "AI agents" work like a goldfish with a keyboard.
You tell them something. They try. They forget. You tell them again. They try again. They forget again.
DeepBrain doesn't work like that.
Here's why that matters 👇
1/10
This is our bet:
Agent frameworks are a commodity. The intelligence behind them — routing, memory, model quality — is the actual product.
DeepBrain is the framework we give away so you can buy the intelligence that actually matters.
9/10
Leaked Gemini 4 Pro benchmarks dont really mean anything.
The true test of an LLMs ability for coding is in the execution of the code.
It either writes clean executable code or it doesnt.
No in between.
Huge if true: Leaked Gemini 4 Pro benchmarks absolutely destroy Astra and Opus 5.5!
What is even going on?
Is @GoogleDeepMind so back?
🤔
Is NVIDIA's Open Agent Safety Platform actually producing any results?
All I hear about is exploits that were obviously preventable from a Controls Engineering standpoint.
Today, with over 100 industry partners, we introduced the NVIDIA Open Agent Safety Platform, bringing together OpenShell and Sentry.
Artificial intelligence is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to come.
But its full promise can only be realized when people have confidence that AI is being built to be safe and deployed with wisdom and responsibility.
This is bigger than a single product. It's the beginning of an open ecosystem to build the trust layer for safe agent systems.
Together, we are building the foundation of the AI economy.
Trust and innovation are not in conflict. Safety is how trust is earned. We must build not only the most capable AI, but the most trusted AI, so that this extraordinary technology can realize its enormous promise for the world. nvda.ws/4hcoq7m
