AI Engineering Tools Landscape

Platform Builders

Prompt → App

"Describe it, get a working app — batteries included"

Best for: Rapid prototypes, MVPs, non-technical builders, PM demos

★

Lovable browser

Full-stack React/TS apps from prompts. Best-in-class UI quality. Lovable Cloud backend (built on Supabase) on by default, or bring your own Supabase. One-click deploy.

~$25/mo Pro (shared across unlimited users) · Credit-based · GitHub sync

Google AI Studio browser

Free Gemini 3.5 access. "Build" mode for vibe coding apps, powered by the Antigravity agent. Deploy to Cloud Run or sync to GitHub. Web and native Android output. Text, image, audio & video models in one playground.

Free tier · Google ecosystem · Gemini API keys

Bolt browser

By StackBlitz. AI builder for websites, apps, and prototypes with design-system support, built-in hosting, databases, auth, analytics, custom domains, and multiple model options.

Free tier · Pro $25/mo · Teams $30/user/mo · Built-in hosting/databases

Replit browser IDE

Cloud IDE + Agent 4. Parallel agents, infinite canvas design, multiple artifacts. Auto-tests and auto-debugs. 50+ languages.

Credit-based · Core $20/mo · Pro $100/mo · Built-in DB, auth & hosting

Emergent browser

Conversational full-stack builder — agents design, code, and deploy web and mobile apps end to end. One-click LLM integration, GitHub sync, custom agents on higher tiers. $70M raise (Jan 2026), ~$50M ARR, 6M+ users.

Free · $17/mo Standard · $167/mo Pro · Enterprise

Figma Make browser

Figma's prompt-to-app tool. Turns a description or an existing Figma design into a working app, with a visual editor over the generated code and live testing. ~60% of Figma's largest customers use it weekly.

Included in Figma plans · Design-to-app · Visual code editing

And many more: v0, Base44, Anything, Blitzy, Hostinger AI Builder, Tempo, Rocket.new …

Common infrastructure defaults

Supabase (Postgres DB, auth, storage) and Vercel (frontend deploy) show up as defaults in most platform builders — but they're not AI-specific. Netlify, Cloudflare Pages, Fly.io, Firebase, or whatever you already use works just as well.

AI-Native Editors

IDE + Agent

"Your editor, supercharged with AI agents"

Best for: Day-to-day dev work, refactoring, feature builds — and devs who still want hands on the keyboard

Cursor desktop IDE

VS Code fork rebuilt around AI, now part of SpaceX (acquired Aug 2026). Indexes entire repos. Composer 2.5 for multi-file edits. Tab autocomplete. Bugbot PR review. Cloud agents. Multi-model incl. Grok 4.6.

$20/mo Pro · $60/mo Pro+ · $200/mo Ultra · $40/user/mo Teams · Also in JetBrains via ACP

GitHub Copilot IDE plugin

Most widely adopted. Agent mode for autonomous edits. Coding Agent assigns issues → PRs. Assign tasks to Claude or Codex agents alongside Copilot's own. MCP + multi-model support.

Free · $10/mo Pro · $39/mo Pro+ · $100/mo Max · $19/user/mo Business · VS Code, JetBrains, Visual Studio, Xcode, Neovim, Eclipse

Google Antigravity desktop IDE + CLI

Google's agent-first IDE. VS Code fork built around orchestrating parallel agents — Editor View plus a Manager Surface. Generates verifiable Artifacts (plans, screenshots, browser recordings). 2.0 adds CLI + SDK. Multi-model: Gemini 3.x, Claude Sonnet/Opus 4.6, gpt-oss.

Free tier (GA) · Gemini 3.x + Claude 4.6 multi-model · IDE + CLI + SDK · Paid via Google AI Pro/Ultra

Kiro desktop IDE + CLI

AWS's spec-driven agentic IDE — turns a prompt into requirements, a design doc, and a task list that live in the repo, then implements against them. IDE + CLI + web + mobile, parallel agents. GA May 2026 as the replacement for Amazon Q Developer. Claude models via Bedrock.

Free tier · Credit-based paid tiers · IDE + CLI + web · Claude via Bedrock

And many more: Devin Desktop (formerly Windsurf), JetBrains AI, Zed AI, Trae …

Bootcamp Focus

Agentic Coding

Agent-First

"Skip the editor — talk directly to the agent"

Best for: Production codebases, deep refactoring, experienced engineers

★

Claude Code CLI + Desktop + Web + IDE

Anthropic's leading agent. Deep codebase understanding via large context. Git-aware. Skills/MCP extensible. Transparent operation. Same engine in terminal, IDE, desktop app, and browser — shared CLAUDE.md, settings, and MCP config.

Claude Fable/Opus/Sonnet · Included in Pro $20/mo, Max $100–200/mo · Configurable approval modes · Local-first

Codex CLI + App

OpenAI's agent. Open-source, Rust-built. Multi-agent workflows (fork sub-tasks). Skills system. Cloud mode for async work. Desktop app + CLI + IDE.

GPT-5.6 Sol/Terra/Luna · Included in ChatGPT Plus $20/mo, Pro $200/mo · Web search built in · Plugin/MCP ecosystem

Gemini CLI CLI

Google's terminal agent. ReAct loop with built-in tools. Strong on frontend/UI tasks. Multimodal input support. Generous free tier.

Gemini 3 Pro/Flash · Free tier (1,000 req/day) · Apache 2.0

OpenCode CLI + IDE + Desktop

Open-source, model-agnostic terminal agent. 75+ LLM providers including local models, plus direct Claude/GPT/Gemini/Copilot sign-in. The answer to "what if I don't want to pick a vendor." ~195K GitHub stars.

Free · 75+ model providers incl. local · Optional Zen curated-model tier · Open source

And many more: Mistral Vibe, Factory Droid CLI, Amp, Kiro CLI, Qwen Code, Crush, Devin, Junie CLI, Aider, Kilo Code, Cline, Warp …

AI Code Review

Automated PR Review

"AI reviewers that catch bugs before your teammates and customers do"

Best for: Teams wanting faster, more consistent code review and fewer bugs in production

Claude Code Review GitHub app

Anthropic's code review agent. Multi-agent analysis on GitHub PRs via per-repo review settings or @claude review. Deep codebase context. Inline comments with severity levels and per-finding reasoning.

Research preview · Team & Enterprise only · ~$15–25/review via usage credits · Not available with Zero Data Retention

GitHub Copilot Code Review GitHub built-in

Native GitHub PR reviewer. Request review from "Copilot" like any teammate. Inline suggestions you can commit directly. Custom coding guidelines support.

Native GitHub review flow · Suggested fixes · Org-level enablement available

Cubic GitHub app

AI-native review platform. 2-click GitHub install. Learns your team's standards. Inline feedback in seconds, one-click fixes, intelligent diff ordering. Also does full codebase scans.

Free for OSS · $40/dev/mo Team · $99/dev/mo Pro · SOC 2 compliant

CodeRabbit GitHub/GitLab app

Most-installed AI app on GitHub (6M+ repos, 17K+ customers). Line-by-line suggestions, 1-click fixes, codegraph analysis. PR + IDE + CLI review across GitHub, GitLab, Azure DevOps & Bitbucket. $143M Series C at $1.5B (Aug 2026).

Free for OSS · $30/dev/mo Essentials · $60 Team · $90 Advanced · Enterprise

Cursor Bugbot GitHub app

Cursor's dedicated PR reviewer. Runs as a background check on GitHub PRs, flags logic bugs with a low false-positive rate, and hands fixes to the Cursor editor or a cloud agent. Customizable via Bugbot Rules. Used as a mandatory pre-merge check at Sentry, Discord, Rippling.

Usage-based (~$1–1.50/review) · Requires Cursor Teams · Owns Graphite

Codex Code Review GitHub app

OpenAI's GitHub PR reviewer. Trigger with @codex review, or auto-review every new PR. Posts a standard GitHub review focused on serious issues. @codex security review (research preview) does a deeper vulnerability pass. Rules live in AGENTS.md.

Included with ChatGPT plans · @codex review · Security review (research preview)

And many more: Devin Review, Greptile, Sourcery, Qodo, Graphite AI Reviews (Cursor), Macroscope, Baz …

AI QA & Testing

Automated Testing

"AI that tests your app so you don't ship bugs"

Best for: QA teams, developers wanting automated test coverage, teams scaling without dedicated QA

Thunders platform

AI test automation platform. Natural language test creation, self-healing tests that adapt to UI changes. Web, mobile, desktop & API testing. AI personas for SEO, accessibility & security audits.

Starter €399/mo · Pro €999/mo · Enterprise custom

Mabl platform

Agentic AI testing platform for web, mobile, APIs, accessibility, and performance. Emphasizes auto-healing, end-to-end coverage, and IDE/CLI workflows.

Free trial · Enterprise pricing · Jira/GitHub/Slack integrations

Ranger platform

AI QA agent that explores your web app, writes and maintains Playwright-style end-to-end tests, runs them as changes land, and triages failures before a human sees them. Built as the QA layer for teams shipping with coding agents. Customers include OpenAI, Suno, Clay.

Free tier · $50/user/mo Growth · Enterprise usage-based

And many more: Momentic, QA Wolf, TestSprite, Spur, Stagehand …

Always-On Agents

Personal AI That Runs 24/7

"AI that works while you sleep — scheduling, automating, and acting on your behalf"

Best for: Personal automation, async workflows, always-available assistants

OpenClaw self-hosted

Open-source personal AI assistant you run on your own devices. Multi-channel, local-first, always-on, with voice, Canvas, memory, skills, and broad messaging-platform support. Now governed by the non-profit OpenClaw Foundation.

Multi-model · MCP/skills · Self-hosted · Free · OpenClaw Foundation

Hermes Agent self-hosted

Open-source always-on agent from Nous Research. Self-improving — writes itself a reusable skill after each task, so it gets more capable over time. Persistent memory, built-in cron scheduler, broad messaging-channel support, runs as a background process on your own infra.

MIT · Self-hosted · Multi-channel · Free

Gemini Spark cloud-hosted

Google's 24/7 personal agent, announced at I/O May 2026. Runs persistently on Google Cloud VMs (works with your laptop closed), takes tasks via Gmail, browses through Chrome, and acts across Gmail, Calendar, Docs, Drive, and MCP-connected apps.

Requires Google AI Ultra $99.99/mo · US-only at launch · Cloud-hosted

And many more: NemoClaw, Claude Cowork, Microsoft Copilot, Lindy, Perplexity Personal Computer …

Orchestrators & Multi-Agent

Orchestrators & Multi-Agent Tools

"One agent is productive — a fleet of them is a different game"

Best for: Engineers who already live in a single-agent CLI and want to run many in parallel, or hand off whole tickets

Single-agent coders like Claude Code are productive; the next leap is running fleets of them. These tools sit above single-agent coders. Loosely mapped to the upper rungs of the Yegge AI-coding maturity model: at stages 6–7 you drive a fleet through an interface and still review and merge the work; at stage 8 and beyond the system runs autonomously from spec to shipped app — the "Dark Factory." This category didn't exist 12 months ago and is consolidating fast — start with the OSS options before paying for anything.

Stage 6–7 · Multi-Agent Interfaces You drive the fleet; you review and merge the work

Symphony OSS · Apache-2.0 · OpenAI

Polls a Linear board, auto-claims tickets, and spawns an isolated Codex session per issue — governed by an in-repo WORKFLOW.md policy file — then returns PRs with proof-of-work. Shipped as an open spec plus an Elixir reference implementation, not a supported product. ~27K stars.

Conductor Proprietary

Polished Mac app (Melty Labs) for running parallel Claude Code, Codex, and Cursor agents in isolated git worktrees — local or cloud — with a central dashboard and inline diff review/merge. Free local tier; Pro $50/mo, Teams $60/user/mo.

Emdash OSS · Apache-2.0

Cross-platform open-source agent dev environment (YC W26). Runs 20+ CLI agents (Claude Code, Codex, Cursor, OpenCode, Amp, Devin, Droid…) in parallel worktrees — local or over SSH — with Linear, Jira, GitHub, GitLab, Asana and more for ticket integration.

Claude Squad OSS · AGPL-3.0

Terminal TUI that manages many agents — Claude Code, Codex, Gemini, Aider, OpenCode, Amp — each in its own tmux session and git worktree. The lightweight, terminal-native way to run a fleet.

Agent Orchestrator (AO) OSS · Apache-2.0

Agent-, runtime-, and tracker-agnostic orchestrator (formerly Composio AO). Each agent gets its own worktree, branch, and PR; when CI fails the agent fixes it, and it answers review comments without per-edit approval.

Orca OSS · MIT

Open-source agent development environment (Stably). Fan one prompt across parallel agents, each in an isolated worktree, compare results, and merge the winner. 30+ CLI agents, GitHub/Linear integration, SSH worktrees, desktop + mobile apps. ~60K stars — the most-starred tool in this category.

And more: Sculptor (Imbue — Docker-isolated Claude Code + Codex agents), Nimbalyst (MIT, successor to Crystal — parallel Claude Code/Codex/OpenCode sessions), cmux (Ghostty-based macOS terminal built for running many agents), Vibe Kanban (Bloop shut down April 2026; repo sunsetting, unmaintained since) …

Stage 8 & Beyond · Autonomous Spec in, app out — the "Dark Factory"

Gas Town OSS · MIT · Steve Yegge

Go-based orchestration running 20–30 parallel Claude Code agents across role-based "crews," paired with the Beads work-tracking primitive. v1.2.1 (June 2026). Self-hosted sessions can run $100–200/hr in API costs; a hosted "Gas Town by Kilo" is now GA.

Cosine (Lumen) Proprietary

Formerly Genie. Now "The Sovereign AI Lab": Lumen Scout/Outpost/Sovereign coding agents for air-gapped and single-tenant enterprise deployment. Lumen Outpost leads its Niche-Bench at 59.3%. Pivoting toward UK-government-backed sovereign model training.

Factory (Droids) Proprietary

Enterprise agent-native development: autonomous "Droids" that own migrations, test generation, and modernization across legacy codebases. $150M Series C at $1.5B (2026); Fortune 500 deployments.

Blitzy Proprietary

Spec-to-app autonomous platform — thousands of agents cooperating for hours to build entire projects. The canonical "Dark Factory." $200M raise at $1.4B (2026). Buy, don't build.