Viking from Jomsborg. The DAO investor. Applied math. Geometric quantization. Building reasoning layer.
jomsborg.eth
Based in Poland
- Tweets 29,107
- Following 7,135
- Followers 1,362
- Likes 61,957
Reasoning Machine treats reasoning as a control loop: observe state, restrict actions with hard invariants, let System 1 choose the next step, invoke System 2 only when needed, verify the result, update state, repeat.
1
33
Rejewski and fellow mathematician-cryptologists Jerzy Różycki and Henryk Zygalski, working at the Polish General Staff's Cipher Bureau, developed techniques and equipment for decrypting the Enigma ciphers.
@elonmusk share it
2
7
345
Invariant Engineering
Invariant Engineering answers: What properties must remain true regardless of which actions are taken?
Control Plane answers: Should this action be allowed?
Execution Plane answers: How do we accomplish the task?
2
21
Since development of MCP-Reasoner server, I try to build a general reasoning control architecture in which complex reasoning strategies are not ad-hoc prompt patterns, but well-formed compositions in a typed categorical framework.
3
2
58
I think there is another layer emerging:
Invariant Engineering
Graphs describe the topology of an AI workflow: states, transitions, dpnd.
But correctness doesn't come from the graph itself. It comes from the properties that must hold regardless of which path execution takes.
2
2
7
1,340
Invariant Engineering (my term) asks a different question:
What properties must remain true throughout the process?
22
Graphs describe the topology of an AI workflow, states, transitions, and dependencies. But correctness doesn't come from the graph itself.
7
The future of AI engineering isn't just smarter agents, but it's controllers that preserve explicit invariants across every state transition.
23
The future of AI engineering isn't just smarter agents. It's controllers that preserve explicit invariants across every state transition.
102
Almost there.
The future of AI engineering isn't just smarter agents, it's controllers that preserve explicit invariants across every state transition.
138
The future of AI engineering isn't just smarter agents. It's controllers that preserve explicit invariants across every state transition.
More on my tweets.
3
The future of AI engineering isn't just smarter agents. It's controllers that preserve explicit invariants across every state transition.
More on my tweets.
2
The future of AI engineering isn't just smarter agents. It's controllers that preserve explicit invariants across every state transition.
More on my tweets.
2
The future of AI engineering isn't just smarter agents. It's controllers that preserve explicit invariants across every state transition.
More on my tweets.
2
The future of AI engineering isn't just smarter agents. It's controllers that preserve explicit invariants across every state transition.
More on my tweets.
10
The future of AI engineering isn't just smarter agents. It's controllers that preserve explicit invariants across every state transition.
More on my tweets.
16
The future of AI engineering isn't just smarter agents—it's controllers that preserve explicit invariants across every state transition.
More on my tweets.
18
The future of AI engineering isn't just smarter agents—it's controllers that preserve explicit invariants across every state transition.
More on my tweets.
2
