Build an agent once, then run it on a phone number or inside your product. Every conversation comes back as a full record: transcript, per-step cost, per-turn latency.
npm install @glytos/node # Node and TypeScript
pip install glytos # Python
go get github.com/Glytos/glytos-sdk-go # Go
composer require glytos/glytos # PHP
dotnet add package Glytos # .NET
npm install @glytos/web # Browser voice and chat
npm install -g @glytos/cli # Command line- Voice and chat, on one engine. Voice agents that take calls on a number you connect or a SIP trunk you bring; chat agents that answer on your site, in your product, or straight from the API. Both are built the same way and observed the same way.
- Official SDKs for Node, Python, Go, PHP, .NET and the browser, plus a CLI, over a public REST API described by OpenAPI.
- Complete session records. Every conversation returns a full transcript with per-step cost and per-turn latency.
- Bring your own providers. Use your own language model, speech, and telephony credentials.
- Global telephony. Multiple carriers out of the box, including local numbers.
- Import what you already built. Agents come across from other platforms with their prompts, model settings and tool schemas, and any agent exports back out as portable, secret-free JSON.
