A Nexus Software Laboratories product
Lyla
Voice-driven cross-platform mobile assistant (serverless, BYO credentials).
You speak, it's transcribed, an agent responds, and the reply is spoken back — with no product-owned backend and no custody of your data.
Product screenshots coming soon.
Why Lyla
No Backend, No Data Custody
Lyla is serverless by design: there is no product-owned backend in the default architecture, and no user data is held anywhere we operate.
Bring Your Own Everything
Your GitHub OAuth, your LLM API key, your speech provider — credentials live in your device keychain, never on a server we operate.
Runs Where You Want
The agent's brain runs on-device, on infrastructure you self-host, or your own cloud provider account — capability is where the brain runs and what tools it reaches, not online vs. offline.
Built for GitHub Work
Talk through GitHub issues, pull requests, and repos by voice, with the agent loop running in-app — nothing orchestrated server-side.
Talk through your GitHub work
Talk through GitHub issues, pull requests, and repos by voice, with the agent loop running in-app — nothing orchestrated server-side.
Voice Capture
Speak it. Lyla links it to the commit.
No backend in between
Lyla is serverless by design: there is no product-owned backend in the default architecture, and no user data is held anywhere we operate.
Serverless
Your device talks directly to GitHub and your LLM.
Your credentials, your device
Your GitHub OAuth, your LLM API key, your speech provider — credentials live in your device keychain, never on a server we operate.
Credentials
Nothing leaves your keychain.