The Lyre Blog
Notes on local AI, wireless computing and what I'm learning while building Lyre.
Keep the project, agents and models on the computer that already has them, then reach in from a phone, another computer or remote desktop. What travels, what stays, and where the limits are.
Read the workflowWhere local inference actually changes the request path, which tradeoffs are real, and how a local-first workspace fits beside the cloud models you already use.
Read the argumentSave desktop apps for Quick Access, link them to useful projects and keep the app beside your agent. A practical workflow for testing changes and giving clearer feedback.
Read the workflow guideConnect provider accounts, choose enabled models and fallback routes, and see which account serves a request. A practical guide to Lyre’s current OmniRoute controls.
Read the explainerChoose local helpers, reviewed work across computers, or a remote model with local project tools. Learn where to configure each mode and how to keep parallel edits separate.
Read the explainerStart with Big Pickle, open a local project, ask your first question and try a small change. Real Lyre screenshots, a starter download and an explanation of what free online AI access includes.
Read the beginner guideSet up Ollama, LM Studio or llama.cpp, download a model and get a local reply. Clear steps for Windows, macOS and Linux, with screenshots and an optional Lyre connection walkthrough.
Read the guideI could see the code. I wanted to use the app. The problems that led me to start building Lyre, from testing on my phone to keeping track of work across computers.
Read the story