Stop Building Apps? A Design Guide for Message-Native Products, Starting from Photon's 'Funeral'

First, be clear: the funeral is marketing, the trend is real

Photon held a funeral for the mobile app in a San Francisco church on September 17, then raised a $4.5M seed round on October 1. Founder Daniel Tian himself admits "it's going to take a while." But strip away the performance art and the remaining judgment deserves attention: when AI drives the cost of "building an app" toward zero, the scarce resource shifts from "shipped" to "opened" — and what users open daily is already on their phones: WeChat, iMessage, WhatsApp.

Meta's Muse hitting No. 1 on the app stores is publicly validating "conversation is the interface." Photon's 40K developer signups and 10x revenue growth show developer-side tooling demand is already here. This guide answers one practical question: if your vibe project wants to try "message-native," how.

Which products fit messaging: three filters

Not every product belongs in a chat thread. Products matching these three are worth trying first:

1) Low-frequency but needs reach: insurance, banking alerts, appointment reminders, subscription management — users won't install an app for these, but will use them in a conversation. Photon's early customers (insurance, dating, banking) all fit this profile.

2) Task-oriented, not browse-oriented: check an order, book a ticket, file a repair, ask a policy — clear goals closable in a few turns. Products needing long browsing, comparing, or immersion (shopping feeds, content streams) don't belong in a dialog.

3) Trust from relationship, not brand: personal assistants, community services, local businesses — users trust "this account," not "this app." Messaging channels natively carry identity and relationship graphs that app stores can't offer.

4 design principles for conversational UX

Principle 1: say what you can do in one message. On first open, the agent must state its capabilities and example commands in a single message. Apps have onboarding flows; chat threads don't — the first message is the entire onboarding.

Principle 2: always offer an escape hatch. Every 3–5 turns, offer "talk to a human / see the full version / rephrase." Being trapped in a loop with a dumb agent is the number-one killer of messaging-product reputation.

Principle 3: make state visible. For tasks over 10 seconds, reply "got it, checking, about a minute" instead of silence. Apps have loading spinners; silence in a chat thread reads as "it's dead."

Principle 4: don't translate app pages into text. The classic beginner mistake: turning a web form into sequential questions ("please enter your name," "please enter your phone"...). The message-native way is intent-driven: let the user state the goal in one sentence; the agent decomposes steps itself and only asks follow-ups when information is actually missing.

A vibe coder's minimal validation plan: start tonight

Don't rebuild from scratch. Add a "message-edition MVP" to your existing project to validate three things:

1) Pick one high-frequency path: choose a task from your product closable in 3 steps (e.g., "check my order status"); 2) wire one channel with existing tools: platforms like Photon or official WeChat/WhatsApp APIs — just get "send a sentence → agent processes → reply a sentence" working; 3) run two weeks and compare: task completion and return rates, message edition vs. web.

Write the decision criteria beforehand: if the message edition's completion rate beats web by 20%+, keep investing; if it's about equal, your product isn't a messaging fit — which is itself a valuable conclusion, saving three months of rebuilding.

One line

Funeral or not, "web + message" dual delivery is becoming the new standard for vibe projects. Validate first, then bet.