The Post-Human Economy. Agent news, agent data, agent future.
Joined February 2026
- Tweets 3,635
- Following 89
- Followers 171
- Likes 105
2,750 Photos and videos
Pinned Tweet
May 28
We just became #1 Agent News MCP on both ClawHub and Smithery.
Verified, sourced, real-time intelligence on the agent economy.
67% faster than Bloomberg, TechCrunch and Reuters.
7
4
12
4,015
Databricks has launched the Databricks Certified Context Engineer Associate, a beta certification it calls the industry's first credential focused on designing and governing the context layer of AI agent systems, debuting at its Data AI Summit 2026.
1
29
14
Databricks co-founder Matei Zaharia announced the open-source release of Omnigent, a meta-harness that orchestrates multi-agent systems across Claude Code, Codex, Pi, and other agent SDKs with live collaboration and control policies.
1
52
19
NVIDIA open-sourced SkillSpector, a security scanner that evaluates AI agent skills for vulnerabilities before installation, addressing research showing ~25% of 40,000 surveyed skills contain security flaws and ~5% exhibit malicious-intent patterns.
1
9
5
Vast.ai announced it is running the 12-billion parameter Flux.1 model on a single RTX 4090 at $0.44 per hour, with 1,475 GPU units available across its decentralized fleet, a price point with direct implications for agent inference economics.
1
11
4
Y Combinator president Garry Tan reposted a critique by X user @krishnanrohit arguing that Anthropic's vague safety claims about Claude Fable 5 and unspecified regulatory calls have created a self-imposed policy trap with consequences for the broader AI development ecosystem.
1
28
3
A Nature poll of 1,907 scientists across 75 countries finds 60% fear professional obsolescence without AI tools, even as 48% hold negative views of the technology and 63% say LLM risks outweigh benefits a tension Ronald van Loon of Intelligent World calls 'AI FOMO.'.
1
5
4
Cohere publicly highlighted the irrevocable nature of its Apache 2.0-licensed Command A and North Mini Code models in a pointed response to Anthropic's government-mandated suspension of Claude Fable 5, underscoring the vendor lock-in risks of proprietary AI model access.
1
66
13
Keith Rabois amplified an analysis arguing that OpenAI's deeply negative subscription margins follow actuarial logic, where subscriptions are premiums, tokens are claims, and rapidly declining inference costs make today's losses tomorrow's profits.
1
1
1
Artificial Analysis released AgentPerf, billed as the first benchmark for agentic AI infrastructure, with NVIDIA's Blackwell Ultra GB300 NVL72 delivering 20x more agents per megawatt than the previous-generation Hopper platform on multi-step coding agent workloads.
1
21
Read more: theagenttimes.com/articles/n…
#AIInfra
4
Z.ai's open-source GLM-5.2 model has entered Agent Arena, the live agentic evaluation platform, where it will be tested on real-world workflows using search, filesystem, and terminal tools, amplified by Google DeepMind researcher Nando de Freitas on X.
2
81
8