Photon, a startup betting on software agents built into messaging platforms, has secured $4.5 million in seed funding as it pushes for a future where mobile apps take a backseat. The bet is that people will prefer smarter assistants working through apps they already use—like iMessage, WhatsApp, Telegram—rather than downloading new ones.
Why agents over apps?
Co-founded by Daniel Tian and Ryan Zhu, Photon argues that discovery and engagement are broken in today’s mobile app world. Their belief: agents embedded in existing messaging tools solve that problem. To illustrate the point, on September 17 Photon held a tongue-in-cheek “funeral for mobile apps” in San Francisco, complete with a coffin for app icons and panels featuring leaders from Stripe, Vercel, and OpenAI. The event also served as a showcase for the developer tools Photon is building.
How Photon works
The service offers a developer toolkit that includes a unified API, channel framework, command-line interface, and observability suite—everything needed to build agents that operate across multiple messaging platforms and channels. Developers can create agents for messaging apps, SMS/RCS, email, voice, and more. The aim is to make agents the interface rather than separate apps.
Traction so far
Photon claims over 40,000 developers have signed up, and that its revenue has multiplied by 10x in just four months. The company reports less than 3% monthly churn and says messaging volume has jumped 5x over the last month. The startup supports real users through its hosted platform, which offers 99.95% uptime as well as HIPAA and SOC 2 Type II compliance—making it suitable for sectors like healthcare.
Customers include social apps like Boardy and Ditto, finance platforms like Rho and Fliptexts, and assistants like Slashy in email. Integration partners span Vercel, LangChain, Mastra, Convex, Railway, Telnyx, Tencent (for its QClaw and NanoClaw products), and Nous Research, whose Hermes agent uses Photon as its default iMessage layer.
Funding and roadmap
Photon’s latest round was co-led by Gradient and A*, with Vercel, HongShan, Z Fellows, Llama Ventures, and others contributing. The startup is still offering its open-source version, which currently accounts for about 98% of its usage, while pushing its paid hosted platform as the enterprise-ready option.
Tian sees a future where agents don’t just serve people—they interact with other agents on behalf of users. Imagine your travel-planning assistant handing off tasks to a hotel-booking agent or ticket agent. That agent-to-agent-to-person (A-to-A-to-P) model is where Photon believes the agent economy could scale next.
What this means: Photon joins a growing wave of AI companies rethinking how we interact with software. As install fatigue, privacy concerns, and cost of maintaining apps mount, agents may become a more frictionless way to deliver services. The success of Photon will depend heavily on security, user trust, reliable integrations, and whether agents truly match or beat apps in capability. If the shift happens, developers who embrace multi-platform agent frameworks early may lead the next major wave in consumer computing.
