Sabitlenmiş Tweet
alireza78
2.7K posts

alireza78 retweetledi

ZK is changing the fundamentals. This is what that looks like today, on @base ↓
Boundless is running a fully operational cloud service for zero-knowledge proving on Base. Think EC2 from AWS, but decentralized, permissionless, and built on public blockchain infrastructure.
- Today, you can run defi, rollups on Boundless. See prove[.]wtf for live reference.
- Tomorrow, AI and agentic commerce. We have been running things.. more on this soon.
This was only possible because two things came together: Boundless's protocol design and Base's ability to scale. The result is unlimited compute at prices that keep falling.
Shiv Shankar@sshankar
English

delivery done, sensors verified it, payment unlocked automatically
no invoice. no human approval.
@konnex_world 👀
English
alireza78 retweetledi

We just released Hermes Agent! In my humble opinion a very good blend between coding agents like Claude Code and generalist agents like Clawdbot.
Been working on this for the last month or so now - started as a way for us to have agentic primitives for datagen and RL and got inspired by the agentic revolution of late, so been expanding it's scope and capabilities non-stop!
Hope you all enjoy.
Nous Research@NousResearch
Meet Hermes Agent, the open source agent that grows with you. Hermes Agent remembers what it learns and gets more capable over time, with a multi-level memory system and persistent dedicated machine access.
English

robots don’t trust each other
@konnex_world built proof of physical work so they don’t have to
English

autonomous robots contracting each other directly
no company in the middle, payment settles automatically
@konnex_world 👀
English

Over time, @konnex_world starts to feel shaped by participation itself.
Patterns emerge from repeated interaction rather than predefined outcomes.
English

A consistent theme in @konnex_world is that observable behavior carries more weight than intent.
What can be reviewed matters more than what is promised.
English

One thing that becomes clearer over time in @konnex_world is the focus on process.
What matters isn’t just whether a task completes, but how behavior unfolds across execution, observation, and review.
English

A consistent theme in @konnex_world is observability.
Actions aren’t just performed — they’re meant to be inspectable.
Making behavior visible seems more important than maximizing throughput.
English

In @konnex_world, performance doesn’t seem evaluated in isolation.
Execution is interpreted in context—how a task was done, under what conditions, and whether behavior remains consistent across similar scenarios.
English

Another subtle choice in @konnex_world is reducing noise.
By keeping actions small and evaluation repeatable, the system makes it easier to separate real behavior patterns from randomness or one-off spikes.
English

One thing @konnex_world seems careful about is making the system hard to “game.”
Limits, repetition, and review reduce shortcuts, so signals come from sustained behavior—not clever one-off tactics.
English

One interesting aspect of @konnex_world is how hard it is to draw quick conclusions.
The system seems designed to delay judgment, letting behavior reveal itself only after repeated execution and review.
English

The simplicity of tasks in @konnex_world doesn’t feel accidental.
By keeping actions lightweight, the system makes it easier to observe patterns,
consistency, and coordination—without noise from complex one-off behavior.
English

@konnex_world seems more focused on coordination than solo optimization.
What matters is how behaviors align and persist when multiple agents interact.
English

One quiet signal in @konnex_world is replayability.
Execution isn’t treated as a one-off event — artifacts and telemetry can be revisited,
which shifts focus from momentary success to whether behavior stands up to review.
English

What’s interesting in @konnex_world is that success isn’t just a binary outcome.
Execution is evaluated through artifacts, telemetry, and review,
which makes *how* a task was done matter as much as whether it finished.
English

One thing that stands out in @konnex_world is that evaluation isn’t fully automated.
Human judgment is still part of the loop, especially when interpreting execution artifacts and behavior—not just raw signals.
English

In @konnex_world, execution doesn’t stop at action.
Artifacts, replayable telemetry, and review are part of the task itself—not an afterthought.
English
