AI in your terminal, editor, design app, and inbox: pick your model

The Lens

Warp went open source. The client landed on GitHub in April 2026 under AGPL v3, with the warpui_core and warpui crates under MIT, and the project now calls itself an agentic development environment rather than a terminal. You can clone it and build it yourself on macOS, Linux, or Windows. What did not open up is the money. Warp's agent runs on credits, and the free tier no longer includes full agent access. Build is $20/mo ($18 annually) for 1,500 credits, Max is $200/mo for 18,000, and Business is $50/user/mo up to 25 seats with SAML SSO, bring your own API keys, and team usage metrics. Enterprise adds admin controls and self-hosted cloud agents. Solo: the free tier is a demo now; budget $20/mo if the agent is why you are here. Small and medium teams: per-seat Build beats Business until SSO matters. Large: Business or Enterprise, where SSO and admin controls live. The catch: the generous free tier was the reason to recommend Warp, and it is not that anymore. An AGPL client does not help when the agent calls a server you cannot run. ghostty-org/ghostty and kovidgoyal/kitty are fully open GPU terminals, no account, no credits.

The Lens

OpenClaw is a self-hosted AI assistant that connects to every chat platform you already use. WhatsApp, Telegram, Slack, Discord, Signal, iMessage, Teams, Matrix, and about 15 more. One local gateway, one brain, every inbox. The setup is real work but the payoff is real too. You run a Node.js daemon on your machine (or a small VPS with Tailscale for always-on). Each messaging channel has its own auth dance: WhatsApp needs phone pairing, Telegram needs a bot token, Slack needs an app. Once wired up, you get voice wake words, browser automation, cron jobs, webhooks, and a skills platform that keeps growing. Solo users: run it on your laptop and bring your own API keys. Power users: put it on a $5 VPS and you have a private AI butler across every platform. There is no paid tier, no cloud service, no data leaving your machine. The catch: "free" still costs money. You need LLM API keys (OpenAI, Anthropic, or local models), and the WhatsApp integration uses an unofficial library that Meta could break tomorrow.

The Lens

Open CoDesign turns text prompts into design artifacts on your desktop. Prompt to React prototype, slide deck, PDF, or marketing asset, all rendered in a sandboxed iframe with a live preview pane. It pitches itself as the open-source Claude Design alternative, MIT-licensed and built around your own model API key. Install in under 90 seconds via Homebrew, winget, or direct download. First launch asks for an API key (Claude, OpenAI, or 20+ providers), paste and go. Everything runs locally; credentials live in ~/.config/open-codesign/config.toml and nothing leaves your machine unless your model provider needs it. Fifteen built-in demos cover landing pages, dashboards, and mobile UIs to get you started. Solo designers and small teams get a real Claude Design substitute without the subscription. You pay model tokens only. Larger teams that need brand consistency or shared component libraries should still look at Figma plus AI plugins, since this is built around individual workflows. The catch: it's young and the Electron app is heavier than a web tool. If you want a polished web design IDE, Claude Design is more refined, but it's $20+ a month and locks you to Anthropic models.

The Lens

Zed is built from scratch in Rust to be the fastest code editor you can run. It is not an Electron app wrapping a browser. Every pixel is GPU-rendered, and it shows: files open instantly, scrolling is smooth, and search across a huge repo feels immediate. Built by the team behind Atom, it has real-time multiplayer editing baked in, so you can share a workspace and pair without setup, plus AI assistant integration with Claude, GPT, and local models through Ollama. After five years in beta, Zed shipped 1.0 in 2026. That matters, because the 'is it still early?' question is now settled. The editor is stable, free, and open source. You can run it forever without paying a cent, using your own API keys for any AI features you want. Pricing is now clear instead of evolving. Personal is free and includes a couple thousand AI edit predictions a month. Pro is 10 dollars a month for unlimited predictions plus 5 dollars of hosted-model token credit, usage-based after that. Business is 30 dollars per seat for org-wide model policies, data governance, and access controls. Solo developers get the fastest editor available for nothing. Small teams gain from the multiplayer editing. Larger teams weigh the governance tier against the one real gap below. The catch is the extension ecosystem. VS Code has tens of thousands of extensions; Zed has hundreds. Before you switch, confirm your language server, your debugger, and the niche extensions your workflow depends on actually exist. The editor is excellent. The ecosystem is still catching up.