Getting started
From the Welcome screen to your first plan. Each step names the button and its shortcut on the Mac.
1
Install
Get ArchitectBuddies™ from the App Store. It runs on iPad with iPadOS 18 or later and on Macs with Apple silicon.
2
Choose a notation
On the Welcome screen, pick a notation card and tap Create First Canvas. Or tap Import / Load Model to open a file.
3
Add elements
Open the Element Selector (⌘E), slide a row to the left and tap Add. Tap the canvas to close the selector.
4
Connect them
Drag one element over another until its edge turns red, and pause. Pick a relationship in the Connector Selector.
5
Arrange the canvas
Tap the canvas title to switch canvases, touch and hold it to choose a diagram type. Grid (⌘G), magnify and undo (⌘Z) are along the top.
6
Set up AI (Pro)
Open the chat with the Architect Buddies™ button (⌘T), then Chat Settings. Decide on Data Security, choose Single Agent (One Buddy) or Team (of Buddies), then select a provider and model. For cloud AI services, paste in the API key obtained from your selected provider.
7
Pick a starting point
An empty chat offers six starting points. ‘Model with One Buddy’ is the quickest first try.
8
Save your model
Save / Export (⌘S) writes a .abmodel bundle and an exchange file for each notation you used.
Running AI in the cloud
Get an API key
Create an account with OpenAI, Anthropic, Google AI Studio, xAI or AI/ML API and generate an API key. Cloud models give the Buddies the best results.
Add it to the app
Open Chat Settings, choose the provider, paste in your key and pick a model. Your model is only sent to the provider you choose. Never share your key with anyone, including us.
Running AI locally
Ollama on your network
Install Ollama on a computer on the same network, choose Ollama in Chat Settings and allow local network access when your iPad asks. Nothing leaves your network.
On the device
Choose the on-device provider and download a model in Chat Settings. Devices with 16 GB of memory or more are recommended. Smaller models are typically not as capable and reliable as larger models.