Private beta · self-hosted · cloud coming

One AI for the whole team.

Luchy joins your team's Discord. Everyone works with it in the same thread, it remembers what you decided and who said it, and it runs on your servers with the model you choose.

# studio › build broken since this morningthread · 3 people · Luchy
M

Mira09:12

The nightly build fails on the shader step. Can you find what changed?

LuchyAPP09:13

It's the shader cache path, changed in yesterday's commit by Tom. I can revert that line and rerun the build.

Remembered“We don't touch main without a review.” — Alex, #studio, 2 Sep

T

Tom09:14

Don't revert, just fix the path. I need the rest of that commit.⏩ taken into account

LuchyAPP09:15

Got it. The fix is ready on a branch, with a review request for Alex.

open pull request · fix/shader-cache-path · 1 fileApproveDeny

Works with the models you already use — Anthropic · OpenAI · Google Gemini · Mistral · Amazon Bedrock · any compatible endpoint

01 — Together

Everyone in the same thread.

No more private chats nobody else sees. The whole team reads, steers mid-task and approves, in the thread where the work happens.

  1. MMira asks
  2. Luchy starts working
  3. TTom corrects the plan. It lands mid-task, no restart
  4. AAlex approves the change

02 — Memory

It remembers what you decided. And who said it.

Decisions, facts and procedures become team memory, each with its source. Anyone can see where a rule comes from, and an admin can revoke it in one click.

/memory rules3 rules

“We don't touch main without a review.”

Alex · #studio · 2 Sep

Revoke

“Builds run at night, never during playtests.”

Mira · #ops · 18 Aug

Revoke

“Release notes go out in French and English.”

Sam · #general · 4 Aug

Revoke

03 — Rhythm

A briefing before you've had coffee.

Every night Luchy tidies its memory and writes up what's active, what changed and what's planned. Schedule your own on top: a stand-up, a Friday retro, a weekly report.

Morning briefing · Mon 08:00

Active — shader pipeline fix, review pending (Alex)

Changed — 14 commits, 2 new rules, playtest moved to Thursday

Planned — release notes in French and English, store page update

Control

Close to your team. Under your control.

On your servers

One private instance per team. Your conversations and memory stay with you. A cloud version is coming.

Nothing risky without a yes

Anything destructive waits for approval from the people your roles allow.

Any model, cost in sight

Bring Anthropic, OpenAI, Gemini, Mistral or your own endpoint, and see what each session costs.

Why Luchy

Not another assistant.

Discord, not Slack

Most AI teammates are built for Slack. Luchy is built for the teams that already work on Discord.

The team, not one person

Personal assistants work for whoever runs them. Luchy works for the team: shared threads, shared memory, roles.

Yours, not rented

Hosted assistants keep your team's knowledge on their side. Luchy keeps it on your servers.

Questions

Where does it run?
On a server you control, one isolated instance per team, deployed with Docker. A hosted cloud version is coming for teams that would rather not run it.
Which models can we use?
Anthropic, OpenAI, Google Gemini, Mistral, Amazon Bedrock, or any compatible endpoint. If one model is saturated, Luchy can switch to a backup.
Do we need Discord?
Discord is where Luchy lives with your team. The same conversations also open from your code editor and through an OpenAI-compatible API, and a web app is on the way.

Give your team its AI teammate.

Luchy is in private beta. Tell us about your team and what you'd want it to do, and we'll write back.

We use this only to answer you. Privacy