Hermeia

Hermeia is a messenger meant to replace Discord for a team: servers with channels, threads, roles with permissions, direct and group messages. A messenger is nothing new in itself - which is exactly why it is hard.

People do not compare it to an abstract ideal but to the Discord and Telegram they use every day and have in their fingers. If the conversation list scrolls with a stutter or opening a channel takes a beat too long, the switch is over within the first week, before anyone appreciates anything else.

The Android and iOS app is built from one codebase, so one team maintains one source of truth instead of two apps drifting apart. And since Hermeia spreads outside the stores, one more job came on top of all this: our own update mechanism.

The first of the two hard problems is smoothness. There is no discount here: a user who scrolls thousands of messages in Telegram every day will feel every stutter in a fraction of a second. The message list has to scroll cleanly on long history, and transitions between screens have to be instant.

This is not cosmetics that would be "nice to have". For a messenger, smoothness is a core feature - as important as the message actually arriving. An app that stutters is perceived as broken, even if it does everything else well.

The second problem is distribution. Hermeia spreads outside the official stores, so we do not have the convenient update channel a store gives for free. Since nobody will push a new version to the user for us, we have to do it ourselves - and in a way that does not make updating painful.

On top of that comes the decision on one codebase for both platforms. Maintaining two separate apps for Android and iOS is double the work and double the room for drift - a fix on one platform that never reached the other. A single codebase with shared logic and data model lifts that weight, while the native finish stays where the platform difference actually shows.

It would be easy to ship a messenger with half the features and add the rest "later". But a team that is to switch from Discord compares one to one: if threads or roles with permissions are missing, they see it at once and go back to the old tool. So from the start we mapped the full model the team genuinely uses.

Servers with channels, threads, roles and permissions, direct and group messages - this is not a selected fragment of Discord but the whole skeleton daily communication rests on. Without it a "replacement" is a replacement in name only.

On start, the client compares its version against what the server advertises, and knows whether the update is optional or required. The manifest has a simple shape: the latest version, the lowest still supported, package addresses and a flag for whether the update is mandatory.

The split into optional and required is the key here. A small fix can wait until the user decides for themselves. But a change that breaks compatibility with the old protocol has to be forced - otherwise part of the team gets stuck on a version that stopped talking to the rest.

Hermeia does not win on a new idea for a messenger - it wins by not lagging behind what people already know on the phone, and by staying under the owner's control instead of depending on a store. This is the quiet kind of quality: it does not catch the eye as long as it works, and it would catch it at once if it did not.

Smoothness and a self-owned update path are two things easiest to put off "for later", and then they come back to bite on the first serious change. We built them from the start, because in a messenger these are not add-ons - they are the foundation everything else stands on.