ai @ northeastern subconscious co-op

Chicago, IL
AI agents connected to all your important apps running 24/7, no Mac Mini needed. Co-Op lets anyone automate their day without knowing how to code. app.subconscious.dev
23
Setting up an AI agent still takes too long. Tool loops, context management, streaming infra. You're a day in before anything works. We fixed it: npx create-subconscious-app my-agent -e vercel-template Full-stack Next.js UI w/ custom tools. One-click Vercel deploy. Try it out!
1
2
3
147
Most dev tools never get a promo video. Too slow, too expensive. I open-sourced an agentic video editor using Subconscious to fix that — describe your product, drop a repo link, get a full video. Free: subconscious-remotion-demo.v…
1
50
Check out this Subconscious open source app I built: it kicks off AI agents that send you personalized emails for free. Pick your topics, set a schedule, wake up to a custom briefing. Sports, stocks, news, research - whatever you want. Repo + live link in comments.
1
1
2
125
Owen Stepan retweeted
we've been working on something kind of crazy. could change the whole agent landscape ngl we're looking for experienced developers to come help give feedback: subconscious.dev/ i'll give you 5 dollars in free credit to test it out dm me for more details @subsysdev
1
3
82
Just built a receipt processing agent. Snap a receipt → AI extracts everything, categorizes based on your spending habits, flags prices as good/bad/fair, stores it in a dashboard. Reducto for document parsing, Subconscious for agent orchestration. Repo + demo below👇
2
1
4
141
Shipped a new CLI and project for Subconscious. Run one npx command, pick an example, and start building immediately. No cloning or setup friction. The new project is an example repo for building real-time, agentic web apps using Subconscious and Convex. Demo below.
1
50
Most resumés are rejected by AI before a human sees them. So I built AI with Subconscious to write resumés for those filters and even the playing field. resume-tailoring-agent.subco…
1
29
Owen Stepan retweeted
AI agents without long-horizon reasoning aren't agents. They're liabilities. Orchestration frameworks collapse under complexity. Real work requires reasoning that spans context, time, and thousands of moving pieces. TIM was built for the chaos.
2
4
229
I don’t think agents should reason without being able to check their work. So I built a CLI agent that runs code while it thinks. Subconscious handles reasoning and context pruning. E2B runs the code in a fast sandbox. Run it with: npx @subcon/e2b-cli
1
2
3
85
Owen Stepan retweeted
Our paper hit #1 on Hugging Face. TIM breaks context limits by treating reasoning as trees, not sequences. The result: unlimited working memory + multi-hop tool use in one inference. Check out the article here: huggingface.co/papers/2507.1…
2
5
94
The only thing stopping agents right now is the context window, mainly because there isn't really a unified way of managing it. You either have to summarize, cut short, write to a readable file, or start a new chat. Switch to Subconscious to have that handled for you.
1
32
AI keeps shipping faster while trust keeps dropping. We are automating creation before we solve verification. That feels backwards.
1
24
Everyone’s talking about AI agents, but most are just scripted workflows. The real shift is agents that reason, persist, and run end to end.
1
22