Best Artificial Intelligence products of the week October 19, 2026

Train with Pasta

A personal trainer powered by Apple Intelligence

Pasta is a personal trainer that lives in your pocket. Ask for a 30-minute upper body workout, and it builds one for you on the spot. During the session, voice guidance keeps you focused, so you can push hard instead of staring at your phone. It also handles longer plans: tell it what you are training for and when you can train, and it writes a plan around those days. The app runs on Apple Intelligence and does not need an account, so your progress and health data never leave your device. You just need an iPhone with Apple Intelligence and iOS 27.

Beam.win

Run AI agents on one Mac and use its windows on another

Keep your agents and builds running on a Mac mini or Mac Studio, then use Beam to bring their Terminal, browser, and other app windows onto the Mac in front of you. Each app appears in its own window with a Dock icon and Command+Tab presence, so moving between your local tasks and remote work feels like one continuous setup. All traffic is encrypted between your Macs, and it is free to try, which makes it easy to test before committing.

Beam

Highly efficient agentic 501B parameter open model

Beam is a sparse Mixture-of-Experts model with 501 billion total parameters, but only 23 billion active during each step. It is built for coding, reasoning, and agentic workloads, so it can handle long, complex tasks while keeping compute and latency practical. The sparse architecture gives you the broad knowledge of a much larger model without paying the full inference cost on every single request.

buddylist

Chat with your coding agents like it's 2005

Buddylist is a free desktop app for macOS on Apple Silicon, Windows, and Linux that puts Claude Code, Codex, Gemini, and OpenCode into an MSN-style buddy list. You can use separate Git worktrees for separate jobs, or group-chat agents in a shared folder. Let them take turns, answer together, or hand unfinished work to another agent. You can review diffs, run terminals, preview your app, and track PR checks right beside the chat. Bring your own agent accounts, and provider access and usage stay separate, so you only pay for the models you already use.

Framefields

Open-source code-first video for AGI era

Framefields is a high-performance, code-first video engine powered by native WebGPU in Node.js. It is built for AI agents and engineers and replaces bulky creative suites with a single npm package. You get motion graphics, Photoshop-grade VFX, 3D scenes, and audio DSP without opening heavy editing software. There is zero Chromium or browser overhead, so you can render deterministic, frame-accurate video and export hardware-accelerated MP4s directly from TypeScript. If your pipeline needs fast, repeatable video generation, this gives you real creative power inside your normal build tools.

RunInfra

Fast open models for your coding agent

Plug Claude Code, Codex, OpenCode, Cline, Droid or Grok Build into fast open models like GLM 5.3, DeepSeek V4.1 Flash and Qwen3.8 27B. Plans from $10 a month, up to 99.5% cache hits so your plan goes further, and zero data retention.

Harness by Autonomous

Every coding agent, every machine, one command center.

Harness is the open-source command center for coding agents. It brings together Claude Code at work, Codex at home, Hermes in the cloud, and every other agent on every machine, side by side. Use the native desktop app, type hn in your terminal, or use the iOS and Android apps to answer agents by voice. Everything is end-to-end encrypted. Then go beyond code with 49 harnesses for CAD, circuits, robots, games, and music. It is free on Mac and Linux.

Hintly for Google Meet™

Real-time hints for every term you miss on Google Meet

Hintly keeps up with your Google Meet calls and explains the terms, names, and topics people mention moments after they come up. You get a quick, clear explanation in the moment, so you never have to pause the conversation or quietly wonder what something means.

Craft Agents

The world's most powerful agents, with the UX they deserve

Craft Agents is fully open source and runs locally on your machine, so your data and API keys never leave it. You can bring any AI provider you like, including Claude, GPT, Gemini, or local models. Agents connect to the tools you already use, such as MCP, APIs, and local files, and they can act across all of them. Code diffs, data tables, diagrams, and live previews show up right inline, so you can review work without switching windows. You can turn results into persistent Pages, teach reusable Skills, and run Automations on a schedule. It all lives in a desktop interface that feels natural to work in.

withHuman

A open-source human approval layer for AI Agents

withHuman is an approval layer for AI agents. It intercepts your agents' tool calls before they run, then applies rules that approve or deny routine actions automatically. Risky actions are sent to the right person on your team for review. It works with Claude Code, Codex, and agents built with LangGraph, Pydantic AI, and more, all through an open protocol called AAP. One set of rules covers every agent your company runs. The project is open source, and a hosted version is available. Use code PRODUCTHUNT to get 3 months of Scale free, valid until Friday.