Devlery - AI news for builders
Devlery blog
AI news for builders.
Thinking Machines Makes AI Collaboration Real Time
Thinking Machines Interaction Models proposes full-duplex collaboration where AI can listen, see, speak, and use tools at the same time.
Claude Agent SDK credits move agent automation into metered budgets
Anthropic is separating Claude Agent SDK usage into monthly credits, making coding-agent automation a budgeted workflow rather than plain subscription usage.
Claude moves into the small business back office
Claude for Small Business packages QuickBooks, PayPal, HubSpot, Canva, and other tools into agentic workflows for SMB operations.
Baidu DAA puts a new metric on the agent era
Baidu proposed Daily Active Agents at Create 2026. The platform race is moving from token consumption toward agents that actually complete work.
Needle brings tool calling down to a 26M on-device model
Cactus Compute Needle is a 26M-parameter local model for tool calling, a small experiment that changes how agent latency, cost, and privacy should be designed.
Honeycomb turns AI agent black boxes into timelines
Honeycomb Agent Observability shows that the production bottleneck for AI agents is moving from model-call logs to handoffs, tool calls, costs, and failure reconstruction.
NVIDIA Is Targeting the RL Training Loop
NVIDIA and Ineffable Intelligence are pointing the model race toward RL infrastructure for agents that learn from experience.
Mistral SDK compromise shows trusted CI can ship malware
The Mini Shai-Hulud attack hit Mistral AI SDK and TanStack packages, exposing a new supply-chain risk around CI, cache poisoning, and OIDC publishing.
OpenAI Realtime 2 turns voice agents into tool callers
OpenAI GPT-Realtime-2 moves voice AI from a conversational interface into a tool-calling agent runtime.
Red Hat is turning Ansible into the agent execution layer
Red Hat Summit 2026 shows how enterprise AI agents may need execution, observability, sandboxing, and governance before they can touch infrastructure.
Frontier AI predeployment review is becoming the new launch gate
CAISI is expanding predeployment evaluation work with Google DeepMind, Microsoft, and xAI, moving frontier AI launches beyond public benchmarks.
Gemini Intelligence turns Android apps into agent tools
Google Gemini Intelligence and AppFunctions move Android apps from screen-first workflows toward local tools that agents can discover and call.