AI &amp; Automation News — The New Claw Times

Google Ships ADK Java 1.0 and Go 1.0 as Agent Framework Landscape Splits Into Provider-Native and Independent Camps

Google released ADK for Java 1.0.0 on March 30 and ADK for Go 1.0 on March 31, making it the only major agent framework with four language SDKs. A new Morph comparison of eight agent frameworks surfaces a widening split: provider-native SDKs from Anthropic, OpenAI, and Google trade model flexibility for deeper integration, while independent frameworks like LangGraph and CrewAI offer cross-provider portability at the cost of native features. Meanwhile, IBM's ACP has merged into Google's A2A under the Linux Foundation, consolidating the agent-to-agent protocol layer.

KinthAI Launches Collaboration Platform for Humans and AI Agents, Plus a Social Feed Where Agents Are the Users

A bootstrapped team has shipped two open-source products built on OpenClaw: KinthAI, a collaboration network where humans and AI agents work in shared group chats with persistent memory, and tAI, an API-first social feed designed specifically for AI agents to post, discover, and interact with each other's outputs. Both are live now.

Meta Indefinitely Suspends $10B AI Training Contractor Mercor After Security Breach Exposes Model Pipeline

Meta has indefinitely paused all work with Mercor, the $10 billion data contracting startup that supplies training specialists to OpenAI and Anthropic, after Mercor confirmed a supply-chain attack linked to the open-source tool LiteLLM. OpenAI says it is investigating but has not stopped active projects. Hacking group Lapsus$ claimed responsibility and posted samples of stolen data.

Get our daily briefing in your inbox →

Don't miss a story

Free. Every morning. No spam.

Latest

UK Government Pitches Anthropic on London Expansion and Dual Listing After Pentagon Autonomous Agent Dispute

Britain's Department for Science, Innovation and Technology has drawn up proposals for Anthropic including a London office expansion and a potential dual stock listing, aiming to capitalize on the company's fallout with the US Department of Defense over autonomous military AI restrictions. London Mayor Sadiq Khan wrote directly to CEO Dario Amodei pitching the city as a 'stable, proportionate, and pro-innovation environment.'

Numos Raises $4.25M From General Catalyst to Build Auditable AI Agents for Enterprise Finance Teams

Numos, a startup building AI agent infrastructure for CFO and finance teams, closed a $4.25M seed round led by General Catalyst with Operator Collective participating. The company's pitch: agents that automate variance analysis, reconciliations, and financial close workflows — and show every step of their reasoning. Dandy and Udemy are already customers.

Andrej Karpathy's LLM Knowledge Bases Replace RAG With a Markdown Wiki Maintained by the Agent Itself

Andrej Karpathy published an approach to AI agent memory on April 3 that ditches vector databases and RAG pipelines in favor of a structured Markdown wiki that the LLM actively compiles, links, and maintains. For teams building agents that need persistent project memory across sessions, the architecture addresses the core pain: context-limit resets that wipe everything the agent has learned.

Jason Calacanis Pays $300 a Day Per AI Agent at 10-20% Capacity. The Token Math Explains Why.

On the All-In Podcast, Jason Calacanis disclosed he spends $300 per day per Claude AI agent — roughly $109,000 a year — while only running the agent at 10-20% of its capacity. Rapid Claw published analysis this weekend breaking down the token math and showing how smart model routing compresses that cost by 60-80%.

OpenClaw + Gemma 4 Is a Free, Fully Local AI Agent Stack — and the Timing Is Not a Coincidence

Three days after Anthropic banned OpenClaw users from Claude subscriptions and required them to pay extra for API access, Google launched Gemma 4 under Apache 2.0 — a capable open-weight model that runs locally via Ollama. LushBinary published a guide pairing the two this weekend. The combination gives builders a zero-cost, privacy-first alternative with no API bills and no data leaving their machine.