Photon has raised a $4.5 million seed round to expand infrastructure for AI agents that reach people through messaging services rather than requiring another standalone app. TechCrunch reported that the round was co-led by Gradient and A*, with participation from Vercel, HSG, Z Fellows, Llama Ventures, Karman and individual investors. The funding follows a deliberately theatrical launch event: on September 17, Photon staged a funeral for mobile apps in a San Francisco church, complete with a coffin for app icons.

The ceremony was partly a joke, but the company’s argument is serious. Co-founder and chief executive Daniel Tian told TechCrunch that a broad move from apps to agents will take time, while describing the direction as inevitable. Photon’s founders believe the obstacle for many new consumer products is no longer simply building them; it is persuading people to discover and install yet another application. Their alternative is to put an agent where users already communicate.

A unified gateway connects messaging, email, voice, and modular AI services.
Photon’s developer stack routes one agent experience across multiple communication channels.

Photon packages that idea as developer infrastructure. According to TechCrunch, its offering combines a unified application programming interface, an extensible framework for communication channels, command-line tools and an observability suite. The system supports iMessage, WhatsApp, Telegram, SMS and RCS, email, voice and other channels. In practice, a developer can build an agent once and make it available across several familiar modes of communication.

The project grew from the founders’ own difficulty distributing consumer apps. Tian and co-founder and chief technology officer Ryan Zhu had built products as students and at hackathons, but struggled with discovery. Tian later created a bot that could respond to friends over iMessage as if it were him. After the underlying technology was released on GitHub, thousands of developers adopted it, TechCrunch reported. Photon says the open-source version still represents 98% of overall usage.

In April, the founders added a hosted subscription service alongside the open-source software. Photon claims the managed version provides 99.95% uptime and meets SOC 2 Type II and HIPAA compliance requirements. Developers can begin on a free tier, while deployments serving more than 10 users must move to one of three paid plans. Those operational and compliance claims come from the company and were not independently tested in the report.

An open-source agent workshop expands into a managed and protected cloud platform.
Photon built a hosted service alongside its heavily used open-source project.

Photon says more than 40,000 developers have signed up, with paying customers including Corgi Insurance, Boardy, Ditto, Rho, Fliptexts and Slashy. It does not disclose revenue, but told TechCrunch that revenue increased tenfold over four months, churn remained below 3%, and messaging volume rose fivefold in the preceding month. Those are self-reported performance figures, so they indicate momentum claimed by the startup rather than independently verified results.

The company is also positioning itself as a connective layer for a wider agent ecosystem. TechCrunch identified integrations or partnerships involving Vercel’s Eve and ChatSDK, Nous Research’s Hermes, LangChain, Mastra, Convex, Render, Railway and Telnyx. Photon also sits beneath Tencent’s QClaw and NanoClaw, according to the report. Tian estimates that the hosted and open-source products together now reach millions of end users, another company-supplied figure.

Photon’s longer-term thesis extends beyond agents talking to people. Tian described a future in which a person’s primary assistant can call specialized agents during a task, such as bringing flight-booking and hotel services into a travel-planning conversation. The company’s financing and developer adoption give it room to test that model, but they do not establish that mobile apps are actually disappearing. For now, Photon is making a narrower bet: messaging can become a distribution layer for AI services that users do not have to seek out in an app store.