Most AI models today sit within a new browser tab or custom mobile application but this has begun to change. OpenClaw, Poke, and Instinct have successively demonstrated the opportunity available by meeting the user where they exist already – messaging channels including iMessage, WhatsApp, SMS and Email. Managing identity, state, deliverability, and analytics across these distinct channels presents a new challenge to startups and enterprises. Daniel Tian and Ryan Zhu ran into this problem last year when building personal agents of their own and decided shortly after to start Photon. Photon now has more than 50,000 developers building on their platform with millions of messages relayed a day.

Messaging channels traditionally built for humans hold two key advantages as a deployment target for developers. The first is the acquisition and onboarding process that is available across a large set of users. WhatsApp has more than 3 billion users, and iMessage ships by default on more than 1 billion active iPhones. Acquiring users is as simple as having them send a text and there is no need to teach users how to use the interface. Secondly, these channels have over the years added rich functionality that typically exceeds those presented by first party applications. Group chats let an agent join a conversation among several people. Tapbacks are a feedback mechanism and reply threads keep context pinned to the right message. Rich media, polls, message effects, and live location are native. iMessage apps and WhatsApp Flows render mini-apps inside a thread so an agent can present a user a custom UI instead of blocks of text. Lastly, payment rails are already in place. With Photon, developers can reach users across these channels while focusing on model development and the user experience rather than infrastructure.

Photon has championed the developer in every step of their journey and have made the experience available wherever you may be building. We’re proud to support their mission and co-lead their $4.5 million seed round with A* alongside participation from Vercel, HongShan, Z Fellows, and more.