daily notes

previous notes →

Themes

Agent coordination became a native interface

Claude Code turned cross-session handoffs into a built-in text channel, while Prime Intellect made arbitrary agent interactions trainable in its RL stack. The common shift is from manually orchestrated agents to first-class coordination primitives.

@ClaudeDevsoriginal post ↗

New in Claude Code: your sessions can now message each other. Instead of having to re-explain yourself in another session, you can now tell Claude to do it. It sends a summary (not your history or files), and the other session picks it up mid-task.

discussion1 selected reply
@ClaudeDevsreply ↗

It goes both ways: you can also ask another session a question and get the answer back in your active session. Claude can also message another session on its own, e.g. when a change it just made affects what that session is working on.

@PrimeIntellectoriginal post ↗

Today, we’re extending our RL stack beyond individual agents to multi-agent systems. You can now express arbitrary agent interactions and train them. https://www.primeintellect.ai/blog/multi-agent-systems

discussion2 selected replies
@PrimeIntellectreply ↗

The core idea: an environment is just a program over agents. Run them sequentially, in parallel, or interleave them turn by turn. Mix models, harnesses, and runtimes. Multi-Agent is a natural extension of verifiers v1. https://t.co/YcLSMSTyxr

@PrimeIntellectreply ↗

Our new abstractions open up a wide range of training setups: - Agentic Judging — solver traces are graded by a judge - Self-Play — a model playing against itself - User-Sim — a user agent interacting with an assistant https://t.co/QZOaDPcEpX

Claude Code cross-session messaging documentation · Prime Intellect multi-agent systems

Must read

Auto mode put a classifier ahead of prompt fatigueClaude

In a controlled study of 1,053 paid testers, human review caught 13.6% of dangerous commands versus 89% for auto mode; the classifier becomes the default for Pro, Max, and Team users on August 14.

Signals

No single-source signals cleared the bar today.

Sentiment

Coordination became a product primitive +0.34

14 posts · 82% confidence

Themes

No themes have been published for this stream yet.

Must read

Signals

China's July CPI cooled to 0.5%National Bureau of Statistics of China

The official July release put CPI at +0.5% year over year and -0.1% month over month, with January–July averaging +0.9%; food, tobacco and catering subtracted 0.22 percentage points.

Public software showed resilience through 1H26@SouthernValue95
@SouthernValue95original post ↗

Software Disruption Analysis: Industry is More Resilient than Feared I studied a cohort of ~100 public software companies from 2022-1H26, and analyzed them for evidence of being AI winners, losers, or resilient. Full list in last tweet. High level takeaways: - 3 years in, there is not broad evidence of AI disruption in software. ARR growth growth remains healthy and stable, growing mid-teens in 1H26. - It is clear which categories of spend are being categorized for growth, maintained, or being de-prioritized. - A cohort of AI winners is growing faster than AI Ad Tech ($GOOG $META $AMZN $APP), but slower than hyperscalers and labs. - The vast majority of public software companies are demonstrating resilience to AI. - The money is to be made buying (or shorting) stocks that are put into the wrong category: believed losers that are actually resilient, or resilient companies that become winners, or shorting winners that are ultimately challenged, etc.

discussion2 selected replies
@SouthernValue95reply ↗

I define winners as companies with stable growth >25% or which have accelerated off a healthy base; resilient companies show stability; losers show rapid deceleration. Below, the NNARR of winners is accelerating, for resilient companies is stable, and for losers is declining. Similarly, organic growth rates for AI winners are accelerating, for resilient companies are stable, for losers are decelerating.

@SouthernValue95reply ↗

Where budgets are flowing is clear: Budgets are growing (34 companies, 15 accelerating): - Infra and network software - Cybersecurity - AI automation and agentic workflow Budgets are stable but not growing (53 businesses, 6 accelerating): - Vertical SaaS - SOR (ERP HCM Finance) - Supply chain and commerce - Design, CAD, EDA software Budgets are de-prioritized (22 businesses, 2 accelerating, 15 decelerating): - Front Office: CRM, MarTech - Work mgmt and content - Contact center and comms

A roughly 100-company cohort shows mid-teens ARR growth through 1H26 and little broad disruption three years into the AI cycle; the result is author-run research, not audited industry data.

Sentiment

Earnings breadth offset cooler China inflation +0.09

97 posts · 88% confidence