Photon has raised $4.5 million in seed funding for infrastructure that lets developers deliver AI agents through messaging services rather than asking users to install another application. The financing follows what the company describes as rapid developer adoption and revenue growth. Its pitch is that an existing conversation can become the entry point to software that performs tasks on a user’s behalf.
The round was co-led by Gradient and A*, with participation from Vercel, HSG, Z Fellows, Llama Ventures, Karman and angel investors, according to TechCrunch. Photon says more than 40,000 developers have signed up and revenue increased tenfold in four months. It has not disclosed the underlying revenue figures.
Messaging provides a different distribution route
Co-founders Daniel Tian and Ryan Zhu encountered a recurring problem while building consumer applications as students: finishing the software did not solve the challenge of getting people to discover it. Their work on an iMessage bot suggested a way to reach people inside a service they already used.
When they made the technology open source, thousands of developers adopted it, TechCrunch reports. Tian left the University of Pennsylvania’s M&T programme, while Zhu left high school and later became a visiting student at MIT’s Media Lab. Tian is Photon’s chief executive and Zhu its chief technology officer.
The company’s current tools include a unified API, a framework for extending communication channels, a command-line interface and an observability suite. Developers can use them to operate agents over iMessage, WhatsApp, Telegram, SMS or RCS, email and voice.
That is an infrastructure proposition rather than a single personal assistant. Photon wants to supply the communication layer beneath many different services. It could therefore serve developers even if another company’s assistant becomes the most visible consumer product, Tian argues.
Open-source use and paid infrastructure are different measures
Photon says the open-source version accounts for 98% of its use. In April, the founders introduced a managed platform sold through subscription tiers. Usage beyond ten users requires moving from the free tier into one of three paid tiers, according to the company.
The hosted service offers 99.95% uptime and is described as SOC 2 Type II and HIPAA-compliant. Those claims are presented by the source as characteristics of Photon’s offering; the article does not provide an independent operational or compliance audit. They are part of the company’s case for supporting more demanding applications, including healthcare agents.
Named customers include Corgi Insurance, Boardy, Ditto, Rho, Fliptexts and Slashy. Their activities span insurance, social introductions, dating, business banking, financial assistance and email. That variety illustrates the intended breadth of the communication layer without establishing that every customer uses it in the same way.
Photon also reports churn below 3% and a fivefold increase in messaging volume over the previous month. These figures, together with the revenue-growth claim, suggest momentum on the company’s account. Developer registrations, open-source usage, paid customers and revenue are nevertheless different measures; the reported sign-up total should not be treated as a count of paying clients.
Agent-to-agent communication remains an ambition
The startup has integrations with a range of development platforms and agent frameworks. TechCrunch names Vercel and Nous Research among its partners, while describing Photon as a messaging layer beneath other agent systems. Vercel is also an investor in the new round.
Tian expects the opportunity to extend beyond agents communicating with people. He describes a future in which an assistant handling a request can bring in other specialised agents—for example, separate services dealing with flights and hotels during travel planning. That is a proposed expansion, not evidence that a universal agent-to-agent network already exists.
The company’s September developer event made its argument conspicuous by staging a tongue-in-cheek funeral for mobile apps. The theatrical presentation reflected Tian’s belief that agents will increasingly replace traditional applications, but it did not establish that users or developers have completed that transition.
For the funded business, the immediate proposition is more specific: developers can reach users through familiar messaging channels and operate their agents on shared infrastructure. The seed round supports that work. Whether the infrastructure becomes a broader communication layer between agents will depend on adoption beyond the human-facing applications Photon serves today.
