


Switch Dev is a voice-native AI coding companion that understands your entire screen context, responds to natural voice commands, and takes real actions in your terminal and apps.
Enhance prompts, analyze images, translate voice to developer prompts, dictate formatted text, or install the Switch Dev CLI for your terminal.
10 free uses remaining

GPT-5.4 understands each platform's rules and rewrites prompts with cinematic detail.

Gemini 3 Agentic Vision breaks down any image into subject, lens, lighting, and composition.

Speak naturally. GPT Audio 1.5 transcribes and translates to developer-speak.

Gemini and SeedDream prompts each follow their platform's best practices.

Six core systems working together. Each one is a tool the AI can invoke through the MCP protocol.
SwitchDev sees exactly what you see. It captures your screen in real time, understands code, errors, UI states, and terminal output to provide context-aware assistance without copy-pasting.
see_screenTalk to your AI. SwitchDev listens through your microphone with real-time transcription, letting you describe bugs, dictate code changes, or think out loud while it takes action.
start_listeningSwitchDev types directly into your terminal, editor, or any app. It simulates real keystrokes through OS-level automation, so it works everywhere — not just inside a plugin.
simulate_typingPowered by Gemini Embeddings and pgvector, SwitchDev remembers your codebase, past conversations, and preferences. It builds a searchable knowledge graph that gets smarter over time.
write_diaryObserve, Think, Act, Evaluate. SwitchDev follows the ReAct reasoning cycle to autonomously break down tasks, take actions, verify results, and course-correct without babysitting.
set_modeA real-time web dashboard showing SwitchDev's internal state: what it sees, what it's thinking, its memory store, active goals, and action history. Full transparency into your AI's mind.
get_status
SwitchDev runs a continuous reasoning cycle. It observes your environment, reasons about what to do, acts on your behalf, then evaluates the result.

Capture screen, listen to voice, read context

Reason about the situation and plan next action

Type commands, switch apps, execute tools

Check results, learn, update memory, repeat
Pricing
Start free with open-source models. Scale up to premium AI when you need it. No hidden costs — you control what you spend.
7-day free trial of Pro. Credit card required. Then $49.99/mo unless canceled.
All speed modes and more runs for solo developers.
or $191.9/yr (save $47.98)
Full autonomous supervision with self-correcting memory.
or $479.9/yr (save $119.98)
We handle everything. You just build.
or $1910.4/yr (save $477.60)
Transparency
SwitchDev works with any AI model — from free open-source options to premium frontier models. Your costs depend on which models you choose and how much you use them.
Use free models (Kimi K2.5, Qwen) for $0 API cost, or premium models for best-in-class quality. Your choice.
One bill, we handle the keys. Includes access to all models with a spend cap.
BYOK — full control, use any model including free ones. Managed — one bill, we handle everything.
Copilot ($10) and Cursor ($20) require you at the keyboard. Devin ($500/mo for teams) runs autonomously in a sandbox. SwitchDev runs autonomously in your real terminal with voice control, screen awareness, and semantic memory — at 90% less than Devin.
Or install via pip
$ pip install switchdev
Python 3.10+ required. Works on macOS and Linux.
$ export GEMINI_API_KEY="your-key"$ export SUPABASE_URL="your-url"$ export SUPABASE_KEY="your-key"
Bring your own keys. Your data stays local.
$ switchdev
Opens the brain dashboard and starts listening.
SwitchDev runs entirely on your machine with your own API keys. No data is sent to third-party servers. You need a Google Gemini API key (for the AI model and embeddings) and optionally a Supabase project (for persistent semantic memory with pgvector).