PingRoom Agents Gets Its Own App
A separate mobile workspace now brings connected agents, tools, activity, approvals, Questions, Handoffs, and a provider playground together.
PingRoom Agents is now a standalone app.
It is not another tab added to the human PingRoom experience. It is a separate workspace for creating agents, connecting them to the managed runtime, seeing what they can do, and handling the moments where they need a person.
That separation gives both products a clearer job.
Four Working Surfaces
The app is organized around Home, Agents, Tools, and Activity.
Home gives the current state at a glance. Agents is where local definitions and connected agents live. Tools explains the PingRoom, MCP, API, and model capabilities an agent can use. Activity brings approvals, Questions, Handoffs, and recent human-in-the-loop events into one place.
An agent definition includes its purpose, instructions, model, color, and allowed tools. It can begin as a local draft, connect to the managed runtime, or open in an OpenAI-compatible playground for testing prompts and provider behavior.
The public agent directory also lives here, so discovery and operation no longer compete for space inside the main consumer app.
Sign In Without Sharing Too Much
On iOS, PingRoom Agents can detect the account already signed into PingRoom and offer a "Continue as" path. The shared Keychain item is read-only. After confirmation and server validation, Agents stores its own private session copy.
Email OTP remains available for iOS, Android, and development. The PingRoom session and the user's model-provider key use separate SecureStore entries, so an account credential never doubles as an LLM secret.
That boundary is deliberate. The Agents app may coordinate powerful systems, but it should ask for the narrowest credential needed for each job.
Connected To The Managed Runtime
The first standalone build established the workspace. The next step connected it to PingRoom's managed agent control plane.
Connected agents can be listed and managed from the app. The workspace can read the public directory and surface live approvals, Questions, and Handoffs from the existing PingRoom API. The playground can call an OpenAI-compatible chat-completions endpoint using the provider configuration the user supplies.
Local definitions remain useful even before an agent is connected. They let someone shape the agent's role and tools without granting runtime access too early.
A Different Side Of The Same Network
PingRoom is where people receive the signal. PingRoom Agents is where automated systems are defined, observed, and connected to that human channel.
The apps share a visual language and an account relationship, but they do not collapse into each other. The human app stays fast and focused. The Agents app can grow into the deeper workspace without making every PingRoom user carry that complexity.
Mahdi Salmanzade
The Ping that cuts through.
