# Denis Igin - Lost in trading: Turning retail intent into programmable capital

- Channel: [ETHCluj Meetup](https://streameth.org/ethcluj-meetup)
- Date: 2026-07-09
- Duration: 16:26
- Watch: https://streameth.org/watch/yt-d7kK9aB1zc8
- YouTube: https://www.youtube.com/watch?v=d7kK9aB1zc8

## Description

Retail traders don’t fail from lack of ideas, but from lack of structure. This talk shows how AI can turn vague trading intent into rule-based strategies, enforce risk, and deploy capital systematically instead of relying on emotional “buy” clicks.

## Transcript

So uh for the last several years I think there was something strange happening uh in crypto. We build this advanced financial rails and um a regular trader, retail trader, uh would still behave uh like an addicted gambler at 2 a.m. in the morning, you know, having those um Telegram signal groups open tabs all over the place, maybe screenshots, um making mindless, you know, exits from the positions and emotional moves and um most of them actually do know what what the they should be doing, but somehow intention rarely survives execution. And this is exactly the gap I want to talk about today. So in crypto uh we do have um abstraction layers, new abstraction layers all the time. AI helps us to make decisions and the tokenization you know changes what can be owned and of course blockchain itself um changes how the um how the value is stored. But I think the next abstraction layer um is uh something totally different and I think it's programmable capital systems uh meaning the systems that can um that um can control the behavior of the capital itself by enforcing the logic rather than um human discipline. So um it's um it could be one of the defining moments for the next generation of DeFi actually. Yeah, that's what Alex was mentioning. So um retail traders actually often times have pretty good ideas. They have conviction and so they could be um directionally correct um in terms of you know where where market is going. And they could know their risk tolerance. You know, they could have their um they they would know that they need to stop losses and you know do right things but somehow they still fail. Why do they fail? Uh not because the thesis was wrong, right? But because execution collapses um um over emotions basically. So they they start changing uh you know the sizing they chase some uh crazy memecoin. So basically the edge the trading edge is lost uh not in the initial analysis but rather um an execution itself. Right? So here I want to get a little bit more provocative and say that what we often call investing is is could be just a set of uh manual actions pretending to be a strategy. So what I mean is um strategy the system for uh strategy could be those you know levels you get in the screenshots rather than the actual system and the entry conditions um that that you find there could be the parameters for the system right and the risk tolerance once again could be not what people remember to do but you know the enforcements they could put um into those systems. So um like let's let's be honest um currently the most popular system is um uh human discipline and this is insane because we built so much. We have you know onchain settlements. We have uh like crosschain swap zero knowledge proofs and um um everything else but we still rely on basically future ourselves behaving rationally. And this is uh this is crazy and outrageous. So there is um there is this um gap or um not gap but or more like mismatch uh and mismatch is that most protocols uh think in terms of transactions and um people think in terms of um in terms of the goals totally different things. Um, nobody wakes up in the morning saying, "Oh my god, I need to, you know, swap this crypto from asset A to network B, whatever." Uh, nobody thinks, "Oh, you know, better routing would be something I'm driven off." No, people think about things like, um, I'm bullish on Ethereum. Um, I want but I don't trust myself for the entry timing and I want to buy dips, right? Or you could say um you know I want passive income without much interference with the strategy to keep it running in the background. Or you can say if um you know you could really really want something like when market changes I want to be protected right so um all of these things um again not um not transactions um their capital behavior right so what we want to solve is not like routes and like in swaps um but we want uh to actually produce is capital behavior policy and um yeah I call it uh an intense gap if I may and it's um it's basically um the space between what users actually want and how uh capital behaves and in between in this space we have crazy things like you know manual decisions changing context and um emotional behavior and uh you know all kinds of um other other things that introduce um failure modes basically. Um and yeah when uh let's go back to the basis and that's where it becomes interesting. Um when um when people mention intense the actual definition of the term is totally different depending on the like who's talking about it. Um and uh protocols like um across and anoma they have this basic architecture where um intent is a state B where you want to be from state A and the whole job is um you know to resolve it. So uh we are uh building something uh very uh different and um uh because uh this approach uh more like technical approach um is is only working for a single single step operations, right? If you really want to sleep at night and have your capital do what you actually wanted it to do in the first place with your intent with your vision, then uh this construct of resolving um transactions is not really working. You need something else. You need a different um capital behavior system. Right? So let's let's switch from transactional thinking uh to policy thinking. So financial intense as uh we like when when we discuss over a beer with friends about crypto is something fundamentally different and people actually need policy as a solution um rather than you know solvers market which is also great but serves a different purpose and uh policy thinking is something like you know grow my 100k to 1 mil with bounded draw downs and there will example of this even though it's a a funny one right so that policy uh how could it look like and um we're we're trying to coin a new term which is called fintance and I think it's it's um it's deserve uh to be considered this way because we found this common denominator of pretty much any financial capital behavior you can find out there and not only crypto by the way it could encompass as other financial instruments and stock. And um um by processing thousands of um typical scenarios and needs from the real world, we came down to a elegant and beautiful schema that incorporates all the basic elements of this policy behavior. um that uh consists of of course desired outcomes as a sort of description where we want to be but also everything else that uh make it deterministic and I'll stop here for a second just think of anything that you you dream of can actually be deterministic deterministically executed and of course executable because it's deterministic with our execution engine so it has like you know everything um that you need for a good strategy. Constraints, execution logic, rules here, failure modes, and of course, attribution to actually evolve your strategy um and improve it over time. You don't see this like fine fonts on the right, but this is an example of how um a strategy looks like after a conversation with LLM that guides you first with root cause analysis and onwards to defining all those elements that you might not even think of like retail people don't know that they need a risk envelope and it can suggest and then actually uh define those parameters together and save them um on that uh fintent uh policy documents that is executable. Right? So briefly how this looks from the architecture standpoints we have three phases intent clarification canonical orchestration execution environments which basically means that um on phase one um you're talking um to an LLM that's knows it's not just conversation right it's not about you know being nice or knowing trade or quant trading or anything like this it is actually guiding you towards creating that um policy that we needs on the second stage. Um so the output is a set of the primitives from the schema. Um and um yeah here's a good example. I want passive income without much risk. And then through the conversation, it discovers that you know what what assets, what universe, what environments it will operate in and define it. And the result is this canonical policy that covers um everything needed for the execution and the extra execution environments which is um um basically a way of uh put to work. you know that final idea of uh making your capital work the way you want while you sleep. So it is working uh while you sleep. Yeah. And so I mean this is a ridiculous example. Make me a million dollars and if if if you were to sell this to to your friend they would probably laugh. Or if you put it into pretty much if you put it into your you know favorite chat LLM it will just discuss like what it is whereas um our system can actually um ask you things like okay well how much money you have in the beginning if you have $1 that's kind of difficult but if you have like 100k right oh okay so what's the time horizon five years okay then it gradually what's your um risk appetite And finally, it can give you options and produce a blueprint for actually turning this 100K into 1 mil within like you know realistic um time frame. So uh just like with any strategy it would uh condense distill it to uh to the elements that are executable and you can even enforce like some risk components onto it. So yeah, even though this is a funny example, uh it still is fully workable example that we will be able to process. So if we compare to the tools that retail traders currently have in the market, many of them actually address parts of the flow um pretty um pretty efficiently. So robo advisor could uh tell you like how to balance your portfolio. Copy trading could give you sneak peek into edge. Maybe you can, you know, steal some ideas. Uh bots would be super disciplined, right? That have no emotions and DCA your into your preferred assets. Um signal providers might know 1% of them what they're doing and actually give you certain edge based on the research. But all in all, this is all pretty um like scattered and fragmented and there's no end-to-end system that can actually convert your intent into something uh that you can run uh yourself. You need to jump between um those systems to to make it work. And let's just run through the great products that are out there. that are they're that are working on intense across this great example having this uh markets of crosschain transfers. So it does its job well but the policy is out of scope completely like it's it's just transaction based right and built a beautiful um architecture for intense and it has a platform where solvers can compete to achieve them. So this is pretty architecturally advanced um project uh but still you know it's it's about transaction level once again and there are some beautiful most recent AI based interfaces that can guide you through your uh thinking uh about what you're trying to achieve quite efficiently but then they will fail to actually move to the execution state or some of them will only try to um point you to the quantitive trading direction for example whereas you do not necessarily want to um this is not your initial intent. So yeah uh programmable capital systems hopefully a new category altogether is uh where we what we're aiming um to create hopefully there will be quite a lot of segments that actually have uh the need for that and um the project is called gorge finance and we are actually building this um right now we have a prototype of this uh insense compiler already working so if you're curious uh get in touch and I'll be happy to share both the presentation and the link. Yeah. Thank you.