The AI coworker that lives in your Slack.
Open-source, self-hosted, and yours to run. Mention it in a thread; it plans, runs real tasks against your tools, builds and executes code in a sandbox, remembers how your team works, and shows you every step and every cent.
Same idea as a hosted AI coworker, minus the black box.
It finishes what it starts
Every request becomes a tracked task with a full log of every step, tool call, approval, and decision.
It computes instead of guessing
Asked for a dashboard or report, Kortny writes and executes real code in an isolated sandbox, verifies the output, and delivers a file or live preview link.
No surprise bills
Per-task model, token, and cost accounting in the built-in dashboard. Bring your own LLM keys, OpenAI, Anthropic, OpenRouter.
It gets better the longer it's there
Workspace memory, episodic recall, and a knowledge graph built from how your team actually works.
Your bot, your identity
Ship it under your own Slack app name and avatar. Kortny runs the brain; you own the face.
Apache-2.0, Docker Compose, your Postgres
No cloud control plane, no vendor in your data path. Clone it, run it, own the whole stack.
You see every step, pay your provider directly, and nothing leaves your control.
Slack activity in, durable work out.
Kortny turns Slack activity into durable work, runs each task through an LLM and tool loop, and posts the result back to the thread it came from.
Mention
@kortny in a thread, DM, or the ✨ side-panel
Plan
Turns the request into a tracked task
Run
LLM + tool loop; executes code in a sandbox
Verify
Checks output, re-runs on mismatch
Deliver
Posts the result back to the thread it came from
What makes it feel like a colleague
Workspace memory
facts & context that persist
Knowledge graph
people, projects, decisions
Ambient observation
watches channels for work
Approvals
gate every outward action
Running in three commands.
Prerequisites: Docker + Docker Compose, a Slack workspace you can install apps into, and one LLM API key.
git clone https://github.com/matangilabs/kortny
cd kortny
cp .env.example .env # add Slack + LLM keys
docker compose up -d --force-recreate
# then, in Slack:
/invite @your-bot-name
@your-bot-name build me a dashboard from the CSV I just uploadedThe operator dashboard at localhost:8080 has tasks, traces, costs, memory, schedules, skills, integrations, model config, the knowledge graph, and users.