Upgrades that run themselves

Vault is a dual-screen handheld terminal for controlling AI agents and locally hosted LLMs from a purpose-built device. This article explains how the prototype combines an ESP32-S3, OLED, and e-paper display with an OpenClaw agent and gpt-oss-120b served on a LAN host through TensorRT-LLM. It covers the interface architecture that separates live task status from completed output, the Python broker that keeps inference off the device, the hardware failures that shaped the build, and the security limits of remote command execution.

This article walks through migrating a mobile QA agent from a single AI SDK tool loop to Vercel Eve. It shows how Eve’s filesystem-first structure, typed tools, and durable sessions make agent behavior easier to inspect, review, and maintain inside EAS Workflows.

AI agents can turn an app idea into working code in hours. They can also overbuild simple features, miss your architecture, and quietly run up cost. In this article, Hristo Totov breaks down five practical lessons for using AI agents with better prompts, better feedback loops, and better engineering judgment.

Agentic systems may not depend only on larger models. This article examines how swarms of smaller models behave under load and what that means for distributed inference, concurrency, and infrastructure design.

Agents Commander introduces a practical way to manage multi-agent environments directly in the CLI. With a minimal communication protocol, it enables agents to collaborate without complex infrastructure or rigid APIs.

This project shows how to let AI agents control UI state in React apps using Jotai and Flows-AI. By moving logic outside components, agents can autonomously update the UI based on prompts and context to enable smarter, automated workflows.

Flows AI is a lightweight library for building AI agent workflows on top of Vercel AI SDK. It provides a simple, flexible way to orchestrate agents as async functions, with built-in control flow agents for parallel execution and composing flows.

Building AI agent workflows doesn't have to be complex. This practical guide walks through how to use Vercel's AI SDK to create chained, routed, parallel, and evaluator-optimizer workflows step-by-step, with code examples you can try yourself.

Leonel Candido da Silva explains eBay's agentic monitoring workflow: specialized agents, MCP, PagerDuty reports, and plans to learn from past incidents.

David Khourshid explains state machines for AI agents, human approval steps, tracing, and handling invalid model outputs in agent workflows.

Nick Miller explains Cursor's move to SpaceXAI, how Colossus changed model training, and the browser experience behind Grok Bot.

Learn what it takes to build a custom AI-powered chatbot for your company, including the challenges, solutions, performance, and security considerations.

Join Callstack, JetBrains, and Expo in Amsterdam on October 22 for talks on AI agents, UI development, and verification across web and mobile.

How connected state, retrieval, memory, and verification carry coding agents across complex, multi-step workflows. Kewin Wereszczyński and Lech Kalinowski discuss the structures that let agents plan, retrieve context, use tools, and keep working across longer tasks.

Mike Grabowski will explain how to close the full agentic development lifecycle across mobile, web, and native platforms.

See how we built and deployed an Eve agent that tracks React and React Native repositories, filters commit noise, and publishes daily Slack updates.

Join 500+ engineers in Warsaw Sept 17-18 2026 for Agent Conf. Learn agentic development, orchestration, and AI-driven workflows.

Stop fitting your workflow into off-the-rack tools. Learn to build your own AI-powered solutions, including a custom slide generator and a collaborative developer notebook.