Relay Keeper SMS-first AI assistant

Keep the conversation moving

Text your AI like it is already in your contacts.

Relay Keeper turns a Twilio number into a calm, human-feeling AI inbox. Users send a text, Relay Keeper routes it to the model they want, and the reply comes back as a simple SMS.

  • No app install required
  • Stateless MVP with Twilio + DeepSeek
  • Built for quick iteration on Render

One phone number.

One familiar interface.

One fast path from idea to reply.

How it works

Simple architecture, clear handoff points.

01

User sends a text

Twilio receives the inbound SMS on your Relay Keeper phone number and posts it to the backend webhook.

02

Relay Keeper asks the model

The API forwards the message to DeepSeek with a concise assistant prompt and keeps the reply short enough for SMS.

03

The reply comes back as SMS

TwiML sends the answer back through Twilio, so the user stays in a normal text thread instead of a web app.

Why SMS works

Friction disappears when the interface already exists.

Fast adoption

There is no new app to download, no onboarding maze, and no extra account for users to remember.

Human rhythm

SMS feels conversational by default, which makes AI interactions feel more natural for reminders, drafting, and quick support.

Operator-friendly MVP

The current build is intentionally small: one Express server, one webhook, one deployment target, and no database yet.

Ready to expand

This structure leaves room for a richer website, user dashboards, analytics, and model-routing features later.

Launch checklist

From repo to live texting in a few moving parts.

  1. Deploy the app on Render as a Node web service.
  2. Add the DEEPSEEK_API_KEY environment variable.
  3. Point Twilio inbound messaging to /sms.
  4. Text the number and confirm the AI reply lands cleanly.