Photon Raises $4.5M to Build the Agent Layer for Messaging

Photon wants to change how people interact with software — and the startup is making a pretty dramatic statement about it. On September 17, the company held a mock funeral for mobile apps in a San Francisco church, complete with speeches and even a real coffin decorated with app icons.

The message behind the event was simple: Photon believes AI agents could eventually replace many of the apps people use today.

Now, the startup has raised $4.5 million in seed funding to build the infrastructure it believes will power that shift. Photon says it has attracted more than 40,000 developers and increased its revenue by 10x in just four months.

The funding round was co-led by Gradient and A*, with participation from Vercel, HongShan, Z Fellows, Llama Ventures, Karman, and several angel investors.

From Apps to Agents

Photon was founded by CEO Daniel Tian and CTO Ryan Zhu, who became interested in the problem while building consumer apps as students and participating in hackathons.

Their challenge was familiar: creating an app was often easier than getting people to download and use it.

That led them to a different approach. Instead of asking users to install another app, why not bring the software to platforms people already use?

Photon focuses on allowing developers to build AI agents that can communicate through services such as iMessage and WhatsApp, along with Telegram, SMS or RCS, email, voice, and other channels.

Tian said the shift away from traditional apps may take time, but he believes the broader direction is difficult to avoid.

The company’s September event was partly a joke, but it also reflected that long-term vision. Alongside the mock funeral, Photon hosted a developer day featuring panels from companies including Vercel, Stripe, and OpenAI.

Building for Developers

Photon is positioning itself as infrastructure for developers who want to build agent-based experiences across multiple communication platforms.

Its toolkit includes a unified API, extensible channel framework, command-line interface (CLI), and observability suite. Together, these tools are designed to help developers build, monitor, and operate agents across different messaging and communication services.

The company’s technology grew out of an open-source project. Tian had previously built a bot that could respond to his friends through iMessage as if it were him. After Tian and Zhu released the technology on GitHub, thousands of developers began using it.

The response convinced the founders to pursue the idea commercially. Tian left UPenn’s M&T program, while Zhu left high school and later became a visiting student at MIT’s Media Lab.

Even today, the open-source version represents 98% of Photon’s usage, according to the company.

A Managed Platform for Businesses

In April, Photon introduced a managed version of its platform and began offering it through subscription plans.

The hosted service provides 99.95% uptime, compared with running the open-source version locally on a Mac. It is also SOC 2 Type II and HIPAA-compliant, allowing Photon to target use cases that require additional security and compliance, including healthcare applications.

Photon says its platform has now passed 40,000 developer sign-ups and has paying customers. Users who need more than 10 users must move from the free plan to one of the company’s three paid tiers.

While Photon isn’t disclosing revenue figures, it says revenue has grown 10x and customer churn remains below 3%. Messaging volume has also increased 5x in the past month.

Its customers include Corgi Insurance, Boardy, Ditto, Rho, Fliptexts, and Slashy, along with other companies developing agents for fintech, consumer AI, and various vertical markets.

Photon has also built integrations with companies and developer platforms including Vercel, Nous Research, LangChain, Mastra, Convex, Render, Railway, and Telnyx.

Nous Research’s Hermes agent uses Photon as its default iMessage layer, while Photon also provides the messaging layer for Tencent’s QClaw and NanoClaw.

The Bigger Opportunity: Agent-to-Agent Communication

Photon’s current focus is connecting AI agents with people, but Tian sees an even larger opportunity ahead.

He believes agents could eventually communicate with other agents on behalf of users.

For example, someone planning a trip could ask an AI assistant to organize the journey. Instead of handling every task itself, that assistant could communicate with specialized agents responsible for booking flights, finding hotels, or completing other parts of the trip.

Tian describes this as an agent-to-agent-to-person, or A-to-A-to-P, model.

Photon estimates that its hosted and open-source technology already reaches millions of end users.

For now, the company is focused on making it easier for developers to bring AI agents into the messaging platforms people already use. But its larger vision goes beyond replacing individual apps: Photon wants to become part of the communication infrastructure connecting people, agents, and eventually other agents.

Also read : Google Alumni Raise $11.3M to Back Enterprise AI Startups