A small tour · 90 seconds
From a prompt
to work done.
Andromeda is a local AI assistant that can reason, use tools, and keep working through a task. Watch one request move from your terminal to a useful result.
01 / Ask Andromedathinking
● ● ●terminal · ~/my-project
>_ Andromeda (local assistant)
model: gpt-oss:20b /model to change
directory: ~/my-project
memory: local /status for details
›Research FastAPI deployment guidance, then write a concise guide in docs/api-guide.md.
✢ Pondering… · 4s · 180 tok/s · esc to interrupt · thinking
Sample run · outputs are illustrativeLocal model required to run your own prompts
THE ANDROMEDA LOOPYour workspace gives the request context
01 · INPUTYour requestPlain language
02 · LOCALOllama modelPrompt + project context
03 · ACTIONTools, when neededSearch · read · write
04 · RESULTUseful answerWith work to review
Tool results go back to the model before it responds.
TRY THE BUILDING BLOCKS
Explore what Andromeda can do.
Each tab opens a small, hands-on preview. Local-only features are simulated here; the links explain how to run them in your own setup.
Local Ollama models
Read the setup guidePick a model name and copy the command to pull it into Ollama.
ollama pull gpt-oss:20bChoose a model to see the Ollama command. This demo does not connect to an Ollama server.
Use ← → to explore · Space to play / pause