// launch update in progress
Run AI agent workflows from Telegram and a mobile app preview, while every agent, CLI, finding, and project lesson syncs back into a shared operational memory layer.
Join the founders waitlist. Get launch updates, early access to the refreshed PhoneAgents.io preview, and the first founders offer when it opens.
Join the founders waitlistThe Agent Stack
PAM is the planning and allocation layer for PhoneAgents: it dispatches specialist agent streams, tracks approvals, monitors progress, and keeps work moving across model providers without losing operational context.
pam dispatch --parallel --approval-gated
Planning & Allocation Manager
Plans waves, assigns specialist lanes, tracks progress, and routes work between Claude, Codex, Pi, and Hermes while preserving human approval gates.
claude --session-id=w53
Primary Coding Agent
Owns broad implementation streams with long-context architectural awareness.
codex run --review
OpenAI Code Agent
Runs implementation and second-opinion review streams in parallel.
pi --fallback --bedrock
Fallback Coding Lane
Keeps work moving when limits, model availability, or routing constraints appear.
hermes daemon --schedule
Orchestration Backbone
Schedules jobs, monitors wave progress, and keeps memory and status loops alive.
Each lane stays inspectable as a terminal card: current status, command, role, and operating metadata are visible before a human approves sensitive writes, pushes, or deployments.