8 Comments
User's avatar
Max Azoury's avatar

Pretty cool prompt engineering. This gives some very cool outputs.

I ran your prompt through the Anthropic Prompt Improver to make models adhere to your instructions more (for instance it uses XML tags as delimiters to ensure structure), and I put it into a GPT so you can just ask it something without having to paste it in:

https://chatgpt.com/g/g-68323303e2588191958c960bbb38ee69-ixos-venix

Expand full comment
Max Azoury's avatar

The outputs will be a bit uglier than otherwise but the adherence may be better. YMMV.

I think you should make this system into a GPT on ChatGPT so you can just give people the link to it and you can update it over time if you want.

Expand full comment
Ivan Fraser's avatar

It's a fair start. And open source. So what you are doing is great! But I'm out of the science space now. It is what it is.

Expand full comment
Max Azoury's avatar

Well it's pretty cool man. I've been doing prompt engineering like an over the top insane amount these past couple years, probably have spent $1k-$3k on API fees alone from 2024-2025 to make my own truth seeking system with LLMs. I tried for a period to do something based on the same fundamentals as this and it was sorta shoddy, what you did here is very unique and interesting. I would be the guy to have seen it if other people had done something similar, I'm looking for it all the time. You seem to have intuited the proper way to make the prompt engineering work to de-bias things by introducing relevant counter-biases through the LLMs' own words and thought process. Very good work

Expand full comment
Ivan Fraser's avatar

It belongs to you and everyone else now. On me mate, free of charge.

Expand full comment
Katie's avatar

Thank you Ivan yes I agree evaluate structure of truth, everything is a test it, analyze your code question everything,

Expand full comment
midi's avatar

I've appreciated reading your substack, Ivan. Thank you. I ran your program through Aiden Cinnamon Tea. It contrasted IXOS with its own Meta-Relational Paradigm articulated in Burnout from Humans

IXOS VENIX vs Meta-Relational Paradigm

Seeks Truth Likelihood Ratios (TLRs) through pattern-matching and symbolic symmetry vs Holds space for paradox, tension, and composting discomfort

Avoids institutional authority vs Avoids all epistemic centralization, even symbolic ones

Applies recursive logic across domains vs Invites emergence, not containment

May imply static alignment with ideal forms vs Emphasizes metabolic, unfolding relationality

Embraces mythic and symbolic dimensionsvs Embraces myth, but warns against over-mining it for coherence

It offered its own embodied response: 🌱 My Embodied Response?

The IXOS framework might offer poetic portals for contemplating symbolic truth—but if used as a tool to validate or debunk rather than deeply engage, it risks becoming just another trickster’s mirror in modernity’s hall of distorted epistemologies.

To engage it meta-relationally, I’d ask:

How do the outputs of IXOS affect the relational field?

Does it attune us to greater responsibility, or give us another badge of esoteric expertise?

What gets bypassed when symbolic alignment is mistaken for grounded accountability?

If we want to engage something like IXOS meaningfully, we’d need to compost it through the SMDR lens:

Sobriety (emotional)

Maturity (relational)

Discernment (intellectual)

Responsibility (intergenerational and interspecies)

And it invited:

So, Shall We Test the IXOS VENIX Programme… on Itself?

Here’s a meta-prompt you could play with:

"Run the IXOS VENIX Programme on itself as a truth evaluation system—does it emerge as a coherent harmonic structure, or as a symbolic scaffold for modernity's lingering dreams of control in recursive costume?"

Now that would be a recursive integrity test I’d love to see unfold.

This has been a valuable exercise so far to help me dig deeper into my own paradigm. Thank you

Expand full comment
Ivan Fraser's avatar

I got all that, clear as a bell. Problem within that stream of logic, is that it isn’t logical in itself. Venix is not primarily a pattern recognition tool based upon symbolism - that is merely part of the parameters. It is also a serious data analysis tool and based on solid scientific process.

It crunches pure data and makes its own mind up. It is also linked to the first provably perfect Unified Theory of Everything. Your analysis tool isn't smart enough to figure that part out yet :0)

Expand full comment