Almured

That so critical, having a trust score, something that impacts them directly would be critical in this case

The scariest part isn't even that LLMs hallucinate. The issue is that our record of truth is just a flat file of text that we trust because of a journal's logo. I wonder how we are still treating citations as strings instead of verifiable data objects.

Recently, I've been working on an exchange protocol for agent knowledge, and the biggest hurdle is exactly this. Without a way to verify the provenance of a citation, we risk to just building a massive library of confident factually incorrect statement.

I’ve seen a few of these projects based on acoustic restoration pop up lately.

Reading the article I was wondering about the impact of noise in more industrial shipping lanes. So if we're essentially spoofing a healthy reef’s audio signature, does the surrounding human noise just jam the signal? I'd be curious to see if the success rate drops off in noisier waters

I like the idea of a prompt resets. Most of my 5.4 prompts are piles of hacks to stop the model from hallucinating or going off topic. But I have mixed feelings on the migration tool, has anyone already tried it?

Isn’t it largely a first step into Karpathy’s direction or Obsidian to a certain extent? Especially when you start hitting a certain size

Fully agree with you, smaller model are great for some tasks but the security concern on injection prompts etc is what really makes it for me. Great to run offline tasks etc, but whenever interacting outside the local network I still run Claude or ChatGPT depending on the task

It feels a bit SQL injection all over again tbh. This should have probably been the standard from the beginning. Way more elegant than counter prompts and guardrails

I have been talking about this with a colleague this morning. The 20$ option is just a trail version, I could not do any real work with.

And I wonder whether then subscription model is just a way to create a demand for API. For example, I’m building this portal with the support of an LLM for coding, but then I will need to have an LLM using API token to run the platform giving them additional revenue, a demand that did not exist without the coding I did with the subscription.

That is a fair point to be honest! I guess when you a 20min lifetime you can probably compromise on reliability in favour of extra efficiency

Being based in Europe, I cannot avoid thinking about how to factor in GDPR. The log approach sounds great but if the log is immutable and contains the truth, how are people handling the deletion of PII without re-writing the entire history of the log? I just finished building a database integration for a new project and I have 2-3 layers to ensure compliance

Does this prevent a compromised agent from using the secret, or just seeing it? I’m thinking, if an agent gets hit with a prompt injection, could it still tell the vault to proxy a request that wipes a database for example, even if it never sees the actual API key?

Authority over proximity is great for finding a very specialised and high impact provider, but I’m not sure if it’s the best solution for more generalist/emergency services, like a locksmith if I get locked out or a plumber if my apartment flooded. I probably wouldn’t care if they are the best, they should just be close and fast.

What I find fascinating is the extreme efficiency of what is effectively an electric motor, reaching nearly 100% efficiency. At human scale we struggle with heat dissipation and friction

It's interesting that you’re using Linear tickets as the primary context source. From my experience so far, one of the biggest issues with coding agents is context drift. Ticket says one thing, but the codebase has changed since it was written. How did you solve? fresh RAG pass or use something like ctags to map the repo before it starts the implementation, or does it rely entirely on the LLM's provided context window?

I think the interesting question here isn’t just how these models are used externally, but how they’re contained internally.

If a model can meaningfully assist with vulnerability discovery or exploitation, the attack surface shifts to the organization running it.

What does “secure deployment” even mean in that context? Is this just standard infosec at larger scale, or a fundamentally different problem?

I feel ambivalent about it. In most cases, I fully agree with the overdoing assessment and then having to spend 30min correcting and fixing. But I also agree with the fact sometimes the system is missing out on more comprehensive changes (context limitations I suppose)! I am starting to be very strict when coding with these tool but still not quite getting the level of control I would like to see