Whelk: Private AI & Agents

An agent that answers to you. Whelk is private AI on hardware you own. Chat with your own models, then arm a conversation with an agent and let it do the work. The record stays on your device, encrypted. YOUR MODELS, IN THE RIGHT ORDER Whelk connects first to Whelk Server, the free server you run on your own Mac or Linux machine. It can also connect straight to a model server you already run, such as LM Studio or Ollama. On an iPhone, iPad or Mac with Apple Intelligence, it can use Apple's on-device model. A cloud provider is used only if you add one, and every answer from it is labeled. CHAT AND WORK, ONE SURFACE - Chat with any model you run. - Arm a conversation with an agent, and it works in place. It asks before anything that reaches past its workspace. - Fork a conversation to try another path. - Dictate on device. Your voice stays on the phone. - On a Mac, open a chat from the menu bar, or send selected text from any app through the Services menu. - Turn on sync to carry your history between your devices, end to end encrypted, through your own Whelk Server. PRIVATE BY WHELK No account. No telemetry. No analytics. Whelk talks only to the servers you choose. Whelk is open core: Whelk Server is Apache-2.0, the apps are source-available, and the source opens at launch. FREE FOR EVERYONE Nothing is metered and nothing is for sale. Enterprise governance is coming.