Photon raises $4.5 million to put AI agents inside messaging apps

Gradient and A* led Photon’s seed round. Its messaging infrastructure connects agents to iMessage, WhatsApp Business and Telegram.

Marcus Reid
October 4, 2026
Industry
2 mins

A person holds a smartphone in both hands while standing beside a window and a plain wall — Illustrative smartphone photograph by Kristin Hardwick, dedicated to the public domain under CC0 1.0

Listen to this article

Photon has raised $4.5 million in seed funding for infrastructure that connects AI agents to messaging apps. Gradient announced the round on October 1, 2026, saying it led the investment alongside A*. Vercel, HongShan and Z Fellows also participated.

Founded by Daniel Tian and Ryan Zhu, Photon addresses the work behind reaching people through existing conversations. Gradient describes the challenge as managing identity, state, message delivery and analytics across different channels.

One agent across several channels

Photon’s Spectrum documentation lists support for iMessage, WhatsApp Business, Telegram and voice calls over SIP using iMessage lines. Developers run one agent server and connect it to each supported interface through providers that handle messages and platform features.

The developer’s server still owns the agent’s behavior, tools, memory, safety and analytics. Photon’s providers handle the interface work. That division leaves product teams responsible for what their agents do once a message arrives.

The channel rules still matter

The official Hermes documentation from Nous Research shows how its agent connects to Photon’s managed iMessage service without a Mac relay. It also documents a practical limit. A shared free line can reply only after the recipient has messaged that line. Sending to a new recipient requires a dedicated line.

Vercel also documents a Photon connector for apps and agents. It supplies credentials when requested at runtime and centralizes their management. Together, these integrations show how messaging infrastructure can fit into an existing agent stack.

For teams evaluating that approach, delivery rules and the division of responsibility deserve as much attention as channel reach. Supporting a messaging interface is one part of making an agent useful in an ongoing conversation.

Texting Woman Phone by Kristin Hardwick, dedicated to the public domain under CC0 1.0. Resized and compressed for illustrative use.

Marcus Reid

Marcus Reid is focused on covering the money, rules, and institutional choices shaping AI. He runs from funding rounds and chip deals to regulation, lawsuits, leadership changes, and the business of building enormous computing systems. Marcus follows the incentives behind the announcement. Who pays, who gains leverage, and what changes for everyone else? The voice is direct, measured, and occasionally dry, especially when a grand promise arrives with very little detail.