Jacob Rafati

AI PhD, CTO & CoFounder of Kredit
Trusted Autonomy for Enterprise AI

[Jacob Rafati @jake__ai]

@zerohedge @grok was it amazon fault or tesla?

[Jacob Rafati @jake__ai]

@bryan_johnson Never

[Jacob Rafati @jake__ai]

@bryan_johnson Happy birthday! You changed my life. My birthday is tomorrow :)

[Jacob Rafati @jake__ai]

@bryan_johnson My birthday is tomorrow. Wow

[Jacob Rafati @jake__ai]

@bryan_johnson Why do you care about comments? Do whatever you want. It’s your freedom. And “people” say whatever they want and that’s their freedom. You can choose not to care.

[Jacob Rafati @jake__ai]

@elonmusk Evolution from pimp to Ironman

[Jacob Rafati @jake__ai]

@NamecheapCEO What is the smell? Is it our mails burnt in server or smell of Cheapname customer churn? My mom always told me Cheap is expensive. I didn't listen to her and there we go losing our customers now bc I saved few bucks on email service.

[Jacob Rafati @jake__ai]

@NamecheapCEO Let's build more data centers in Arizona. Why not?

[Jacob Rafati @jake__ai]

@elonmusk The XAI logo is fkd. Just FYI.

[Jacob Rafati @jake__ai]

@elonmusk There is no Toyota Camry there?

[Jacob Rafati @jake__ai]

@oelma__ Salad

[Jacob Rafati @jake__ai]

Hi folks, can you please uprank my hackernews?
https://t.co/7iEs39rWAu

[Jacob Rafati @jake__ai]

Watch a demo of Kredit.

Kredit is the real-time risk intelligence infrastructure for AI agents. Companies use Kredit to simulate fleet behavior, forecast costs and operational risks before deployment, score consequential actions before execution, and continuously optimize budgets, resources, limits, and autonomy.

Just as credit bureaus made lending scalable, Kredit is on a mission to make trusted autonomy scalable, helping companies safely scale agent operations and unlock a $20 trillion opportunity.

Founders: Reza Jalali, PhD in math, 3x founder and former Global Head of Enterprise Risk at BlackRock; Jake Rafati, PhD in CS, 3x founder, built enterprise AI systems at Expedia and Dell tied to $1 billion in revenue.

[Jacob Rafati @jake__ai]

@Jason A33

[Jacob Rafati @jake__ai]

@alpaysh Where is Larry David?

[Jacob Rafati @jake__ai]

@Adams_Tech_AI Damn, you should have been the $2T IPO. Such a shame!

[Jacob Rafati @jake__ai]

LLM is not equal to AI.

As an AI scientist with 13+ years in AI and ML, I'm shocked that LLMs are called AI as if AI didn't exist before ChatGPT. We had more mature, more controlled AI before LLMs.

Seriously, pay attention: LLMs are trained to predict the next token, not to accomplish your life goals. When you ask one to write a sales pitch, it isn't trying to sell �� it's filling in sentences. That's why 99.99% of AI-generated pitches suck and human pitches are still better.

LLMs don't care about the task. They're writing sentences. That's why their actions carry risk. As a human, you don't accomplish tasks by producing sentences — there are millions of other considerations involved.

We expect LLMs to have our understanding and common sense. They don't. They're a probability distribution over tokens learned from stolen Reddit threads.

The risk is massive and everyone is oblivious to its magnitude. I hear this 100 times a day: "As frontier models get better, their risk decreases."

Dude, no. That's not going to happen.

Yes, guardrails are necessary. No, static guardrails can't catch dynamic, unpredicted risks.

We're building Kredit on classical ML to detect the risks of LLMs. For now, LLMs are the best models we have to empower agents — until we find better. So instead of staying passive, we're building the risk intelligence infrastructure for trusted autonomy. Every agent intent is scored, monitored, audited. Trusted actions clear; risky ones are blocked pre-execution.

We're enabling a $20T industry.

https://t.co/0xXI90c8SU

[Jacob Rafati @jake__ai]

@elonmusk why do you catch some and leave others in ocean?

[Jacob Rafati @jake__ai]

@bryan_johnson I have been working on decoding the code of aging for 2 years now. This post validates that I am not crazy. I have a PhD in AI and I have sorted a plan on super fast drug discovery, in-vitro testing, and ultimately reach biological immortality. Let's chat.

[Jacob Rafati @jake__ai]

@AnthropicAI Please take Fable 5 out. No need.