Software
MailPals
Group newsletter over iMessage.
Group chats are noisy. MailPals turns one into a weekly ritual. A bot privately DMs each member with a handful of prompts — one at a time, like a conversation — and members respond however they want: text, voice notes, photos. At the end of the week, every response is compiled and delivered back to the group as a single newsletter issue. Delivered issues spark reply threads where members react to each other's answers — the newsletter becomes the conversation starter.
This is a mobile-native spinoff of the open source MailPals project, which I contribute to. The original sends prompts via email and collects responses through a web form. This version moves the entire experience into iMessage — no app to download, no links to click. The chat is the product.
The Interview
Each week, the bot sends 3 randomly selected prompts to each member via private DM. The conversation is linear — answer one prompt, get the next. Voice notes are accepted and transcribed automatically. After answering all three, any additional messages become bonus content in the issue.
Animated — the bot interviews each member one prompt at a time
Issue Delivery
At delivery time, responses are compiled by prompt and sent to the group chat. The issue lands as a stream of bot messages — members reply directly to spark conversation threads.
The compiled issue as it appears in the group chat
Prompt Pool
102 seed prompts across 9 categories — from introspective to absurd. Three are randomly selected each week, and the bot never repeats a prompt back-to-back. Members can also submit their own.
Sample prompts rotate automatically across categories
Weekly Lifecycle
Every issue follows a fully automated four-stage cycle. The group admin sets the schedule once — prompt day, delivery day, timezone — and the bot handles the rest, including creating the next issue after each delivery.
Built With
A mobile-native spinoff of the open source MailPals project. Python backend with FastAPI and async SQLAlchemy. iMessage integration via BlueBubbles bridge running on a Mac. Voice notes transcribed with OpenAI Whisper. Real-time test UI over WebSocket for multi-user simulation across browser tabs.