iiinova — where your digital self keeps evolving

olup.bsky.social

@olup.bsky.social

olup.bsky.social @olup.bsky.social · 25/08/2026

From Hacker News today! Headlong is close to what I imagined: an agent thinking on its own, with one stream shared by a team. It can connect people and ongoing work but also blur boundaries between conversations. Fascinating, and with security challenges. www.laude.org/updates/head...

laude.org

Headlong: a microharness for persistent agents // Laude Institute

Self-guided agents that think continuously

000

olup.bsky.social @olup.bsky.social · 23/08/2026

Agreed, and I recognize this is the original post. But is there a way forward, or are those definitive roadblocks? BTW, these are problems with individual humans as well. How did we solve them in the person's training?

000

olup.bsky.social @olup.bsky.social · 23/08/2026

For completeness' sake's - they are out of training LLMs entirely having sold tech and people to Nvidia, and going next to be a neocloud or somthing.

001

olup.bsky.social @olup.bsky.social · 23/08/2026

Today's AI agents are personal and turn-based. They belong to one person, good for accountability, security, and distribution of power. What's next? A company-owned agent that combines signals across people and channels and shapes ongoing tracks of work through a shared memory and execution core?

200

olup.bsky.social @olup.bsky.social · 25/07/2026

I hear Poolside is aiming for bigger goals, but I really wish we'd get an open-source small model champion throwing all its focus into pushing the intelligence-per-parameter ratio to the extreme. Small performant models will be the lasting legacy of this AI era.

100

olup.bsky.social @olup.bsky.social · 25/07/2026

I read a lot of good things about coding with the Laguna Small 2.1 model from @poolsideai.bsky.social. Now, given that you can run this on a DGX Spark with 2000 kWh per year, it is actually very possible to have your own solar-powered inference for coding.

210

olup.bsky.social @olup.bsky.social · 03/07/2026

Felt sane while reading this. blog.okturtles.org/2026/07/shor... AI is a great tool if you know how to use it. Do you know how to use it? As leaders, how do we know our next hire does?

blog.okturtles.org

The Short Leash AI Coding Method For Beating Fable

000

olup.bsky.social @olup.bsky.social · 21/06/2026

So your company, my company, we probably should not build the customer's ai. Model labs build the model, agent labs build the harness. This is the new browser. Your service needs to be integrated. Great api, CLI, Mcp. Secure auth delegation.

000

olup.bsky.social @olup.bsky.social · 21/06/2026

No one wants your ai agent. It's a pain. What users need is a way to plug their agent to your service. It's scary, you lose control. But it's the future. I want to rent my car, find my flight, analyze my data analytics, deploy my stuff, buy my groceries, through my agent.

100

olup.bsky.social @olup.bsky.social · 08/06/2026

Would love to hear about Jam's direction (esp. on concurrency model, closures, and interface support)—not sure how to reach you, though. In any case, really cool work you're taking on.

120

olup.bsky.social @olup.bsky.social · 07/06/2026

I got excited about the Jam (jamlang.org/reference/ , pre 0.1) language from @raphamorim.bsky.social. No-GC memory-safe language that wants to be very easy and fun to learn and use. Like Rust meets Zig (Risg?) I tried it around on a JSON Serde and a toy web server. Will keep following closely.

jamlang.org

Language Reference - Jam

Jam Programming Language

141