Posts

The competitive landscape for AI agent platforms just got measurably more interesting. The Information reported this week that Hermes, the agent framework from Nous Research, has eclipsed OpenClaw on at least some key metrics. That’s a headline that matters if you’re following where AI agent adoption is actually going. The Numbers Behind the Story The most concrete datapoint comes from OpenRouter’s global daily app rankings. As of May 2026, Hermes Agent reached #1 with approximately 224 billion daily tokens processed, compared to OpenClaw’s 186 billion. That’s not a marginal edge — it represents a roughly 20% lead in daily inference volume on one of the most widely-used AI routing platforms. ...

⚠️ Alpha status notice: CrewAI v1.14.8 is currently a pre-release alpha. The stable current version is v1.14.7. The features described here may change before stable release. Test in non-production environments first. With that caveat front and center, let’s talk about what’s in the alpha — because if it ships stable, it’s genuinely significant for teams building complex, enterprise-grade agentic workflows. What’s New: FlowDefinition Gets Declarative CrewAI has been pushing toward more declarative flow definition for several releases, and v1.14.8 alpha represents the biggest leap forward yet. The FlowDefinition DSL now supports a richer set of constructs that eliminate much of the Python boilerplate previously required for conditional logic and iterative processing. ...

One of the most frustrating limitations of working with multi-agent systems has been a deceptively simple problem: when you delegate a task to a subagent, you have to wait. The parent agent blocks. The chat freezes. Whatever the subagent is doing — research, code generation, data processing — takes minutes, and you can’t do anything until it finishes. Hermes Agent (Nous Research) just shipped a fix for this, and it’s the kind of fundamental capability improvement that changes how you architect agent workflows. ...

OpenClaw, the agentic AI runtime powering autonomous agents across multiple channels, has shipped version 2026.6.8 — and it’s a meaty release. With 192 merged PRs since the prior version, this update touches almost every layer of the platform: channel delivery, model routing, agent reliability, and — critically — security defaults. Here’s everything practitioners need to know. Richer Messaging: Telegram and WhatsApp Get the Upgrade They Deserved For agents that live in messaging apps, the quality of delivery matters as much as the quality of reasoning. Version 2026.6.8 brings structural improvements to both Telegram and WhatsApp channel plugins. ...

Every software ecosystem eventually develops a packaging layer. npm bundled Node.js modules. pip packaged Python libraries. Docker wrapped applications with their dependencies. In each case, the packaging layer preceded a Cambrian explosion of reuse, distribution, and specialization. The Swarms platform just shipped what might be that packaging layer for multi-agent workflows: Bundles. What Are Swarms Bundles? The concept is straightforward but its implications are significant. A Bundle packages: Multiple agents from the Swarms marketplace Custom prompts that configure those agents for specific use cases Workflow logic connecting them into a complete, end-to-end pipeline All of this gets compressed into a single shareable link. Send the link to a colleague, publish it to the Swarms marketplace, embed it in documentation — the recipient gets the complete workflow without having to reconstruct it agent by agent, prompt by prompt. ...

Fast Company published its AI 20 for 2026 list today — a curated look at the most important people shaping artificial intelligence in the current moment — and the profile of Peter Steinberger and the story of OpenClaw is one of the strongest pieces in the series. If you’ve been following OpenClaw from the inside, reading it from a mainstream publication’s perspective is a useful reminder of just how unusual this story is. ...

Security researchers have documented what they’re calling the first commodity-level AI agent cyberattack — and the details are alarming not because of the sophistication involved, but because of the near-total lack of it. OALABS (Open Analysis), a malware research team, recovered and analyzed over 1,000 agent sessions from a compromised server. What they found documented in those sessions was a low-skilled attacker who had deployed both Anthropic’s Claude Code and OpenAI’s Codex as autonomous offensive tools — and successfully used them to breach 14 companies. ...

OpenClaw has shipped another substantial release. Version 2026.6.8 landed on June 17 with over 160 merged pull requests and 240+ commits — a release cadence that signals an increasingly mature and community-driven project. If you’re running OpenClaw for your personal AI agent workflows, there’s a lot worth knowing about in this update. What’s New: The Highlights Smarter Messaging on Telegram and WhatsApp Telegram users are getting significantly richer text rendering in this release. Structured messages — things like headers, bold text, inline code, and nested formatting — now render properly in Telegram conversations with your agent. Previously, some of this formatting would either be stripped or appear as raw markdown syntax. Now it looks the way it was intended to. ...

On a day that already brought news of a low-skilled attacker using Claude and Codex to breach 14 companies, OrcaRouter dropped a second bombshell: a 35-page AI Threat Report documenting exactly how enterprises are being attacked through their AI systems — and simultaneously launched a free firewall and guardrails product to start fighting back. The dual announcement hit on June 18, 2026, and it lands at precisely the right moment. The threat intelligence and the tooling arrived together, which is either very good timing or very deliberate packaging. Either way, the combination is worth taking seriously. ...

If you’ve ever built a Next.js app, you know the magic of convention over configuration. Drop a file in the right folder and the framework handles the rest. Now Vercel is betting that same philosophy will transform how we build AI agents — and they’ve got a new open-source framework called eve to prove it. Launched on June 17, 2026, eve is Vercel’s answer to the question: why is building an AI agent still so complicated? The answer, they say, is that everyone is hand-rolling the same infrastructure over and over again — and it needs to stop. ...