Photon Agent Communication

Overview

Photon is presented as a communication layer that lets long-running agents reach users through familiar messaging and voice channels.

Current Profile

The source positions Photon between background execution and meaningful human oversight. An agent can use iMessage, WhatsApp, phone, Slack, Telegram, and planned email to ask about ambiguity, report progress, deliver summaries, or escalate an urgent decision rather than silently guessing or requiring the user to watch a dedicated dashboard.

Its durable value proposition is not another chat front end. It is a control path for Long-Horizon Agent Reliability / 长期智能体可靠性 and Human-in-the-Loop Agent Governance when tasks outlast one session. The source does not establish Photon’s architecture, customers, security model, delivery reliability, pricing, or adoption.

Key Characteristics

• Connects background agents to users through existing communication habits.
• Supports status updates, questions, urgent confirmation, and final handoff.
• Treats text and phone contact as operational escalation channels.
• Sits alongside memory, state, permissions, compliance, and recovery rather than replacing them.
• Depends on correct identity, routing, consent, and urgency handling across third-party platforms.

Evidence

• Channel coverage: 从免费披萨到订错机票：硅谷创业者正在解决 Agent 的哪些麻烦？ | 硅谷线下活动 S10E33 names iMessage, WhatsApp, phone, Slack, Telegram, and planned email.
• Human escalation: 从免费披萨到订错机票：硅谷创业者正在解决 Agent 的哪些麻烦？ | 硅谷线下活动 S10E33 presents user contact as a safeguard when an agent is uncertain or faces a high-impact step.
• Continuing communication: 从免费披萨到订错机票：硅谷创业者正在解决 Agent 的哪些麻烦？ | 硅谷线下活动 S10E33 describes progress queries, result delivery, daily briefings, emergency confirmation, and family-group updates.

Qualifications

The bounded evidence is one episode summary. It does not independently verify the company, product implementation, channel permissions, privacy controls, message delivery, call disclosure, security, commercial traction, or comparative advantage.

What Changed

• Created a source-scoped profile of the agent-to-human communication and escalation layer.

Relationships

• IM Agent Interfaces - interface category Photon extends from intake into escalation and handoff.
• Human-in-the-Loop Agent Governance - governance relationship implemented through timely user contact.
• Long-Horizon Agent Reliability / 长期智能体可靠性 - reliability relationship for resolving ambiguity during extended work.
• Agent Permission Boundaries - authority relationship that determines which decisions require confirmation.
• Agent Identity And Authentication - identity relationship needed to show who the agent represents.