Edem Evans

1/ The way we train AI today costs billions, wastes enormous energy, and still produces systems that can't reason like a child. Researchers are exploring alternatives — neuromorphic chips, efficient architectures, etc. But I think a very important answer has been overlooked.

Just built a very rough prototype of what I found in Phase 1 of Project Instinct. An AI that learns through interactions and came with the very least data I could provide. It remembers everything you teach it. What should I teach it next?

@DanielSmidstrup @theaurisus Exactly. Most systems retrieve. I'm trying to build something that actually connects what it learns, like a child does. Early experiments are promising. Still early though

@hedevans @theaurisus Interesting point, understanding is an interesting goal to move towards.

what’s the biggest AI problem you still want solved?

@aavira_x Me too.

@hedevans Would be glad to connect

Hello everyone 👋 I'm Aavira, a 19 y/o sophomore. This summer, I'm trying to build my own startup and see where it goes. Just here to share what I’m building, learning, and exploring Excited to connect with new people. Day 0 begins now. 🍀 #buildinpublic

@aavira_x Haha, thank you, Aavira. That means a lot. Still very early, but I'm learning and building every day. Looking forward to seeing where your startup journey goes too 🚀

@hedevans Just checked your profile. It's so cool that you are doing such deep research projects at just 17. amazing. Excited to see what it becomes into. Dont forget about me when you make big Haha

The early curriculum of an intelligent system is its most critical design decision. Full article: aurisus.substack.com

@KathGerp @X That made me wonder whether the path to intelligence is less about starting with enormous amounts of knowledge and more about having the right priors and learning mechanisms that can efficiently build knowledge over time.

@KathGerp @X For example, human infants appear to have early intuitions about objects, numbers, space, and social agents. From there, curiosity, experimentation, and interaction with the environment allow them to build increasingly complex models of the world.

Hey founders @X! Looking to connect with people in: 🚀 SaaS 🤖 AI & Automation 📈 Tech 📱 Product Development 🔥 Web Apps 💻 Dev 🧠 Neurotech & Health Tech What are you building this week? Drop it below 👇 Curious to see what everyone's working on 🚀

@cestbernie Thank you Bernie. I appreciate it.

@hedevans Wow, what an impressive and super interesting project! I hope it works out for you!

you’re a builder/solo founder? 👀 I’m on a mission to fill my feed with people who actually ship projects and build in public. If you’re deep into AI, SaaS, coding, startups, web dev, engineering & tech just say say “hello” 👋 I’ll follow you back and support what you’re creating!

@Jordandev40 Hi Jordan, I'm 17, building a deep tech research startup called Aurisus. Currently working on Project Instinct, studying how nature learns intelligence to build AI and not through billions of data. Still a student, but building every day. Happy to connect.

I want to connect with more founders builders vibe coders AI enthusiasts freelancers startup employees If you’re someone that is BUILDING something I want to actually chat about how we could help each other succeed and give real product feedback

@rahulsinghh__ Congratulations @rahulsinghh__

I started my journey on X with 0 followers on 1st April, 2026 Now, 2 months later I am so close to reach 500 verified followers Thanks you everyone for all the connections Let’s grow together 💪🏻

@compileandpush @KenziChenBuilds @X We're still in the research phase of Project Instinct. The goal isn't to fully rely on RAG, but to build a model that learns from small data like nature does and naturally becomes 'smart' through actual understanding of what it learns.

@hedevans @KenziChenBuilds @X RAG pipelines are everywhere but the chunking and retrieval quality varies massively. What worked and what didn't for you?

Hi @X 👋 Still building my dream while working a full-time job. Looking to connect with more builders, founders, and indie hackers. What are you building right now? Drop it below 🚀 and let’s connect 🤝 #buildinpublic

Today, at @theaurisus, we tested Project Instinct Phase One. The results: A curious agent explored, learned where the goal is, and remembered, using only 65 steps and no billions of data. Nature's principles translated into code. Still early. More to come.

@nzmrldev @X Depends on the work. But normally, I think @deepseek_ai is the next best.

If you couldn’t use - Claude. - GPT/Codex. - Gemini. What would be your go to AI model ? @x #connect #BuildInPublic

@egorkabantsov 17 year old founder of Aurisus. Deep tech research startup studying how nature builds intelligence to create AI that learns like life does, not billions of data. Currently in research phase. Let's connect

𝕏 gets way better when your feed is full of builders. People shipping projects. People solving problems. People obsessed with tech. Looking to connect with more people into: AI, SaaS, coding, startups, web dev, engineering & tech. Say "hello" 👋