An AI team of your own. For everyone.

English · 简体中文

Website · Documentation · Resource market · Issues

Teloa is an AI-Native Team Studio for building, collaborating with, and running an AI team. Teloa Free is its self-hosted personal edition. Give each AI employee a role, relevant knowledge, skills and tools. Assign work, follow its progress, and keep the results connected to the task and business records they came from.

You define the goals, access and approval rules. Employees work within those boundaries, while you handle decisions and review deliverables. Teloa uses DeepSeek Harness (DSH) as its agent runtime.

Free edition: one human user, multiple AI employees, running on your computer through a Web interface. This repository provides the Free source code under Apache-2.0. The current source version is 0.2.0-alpha.7.

Product interface with example employees and work. A fresh installation does not require this demo data.

A business organizes records, people and resources. An AI employee owns a role; a skill supplies a method; knowledge provides evidence; a connector gives access to a system. A task tracks an assignment, a deliverable preserves its output, and a dashboard summarizes the business records.

flowchart LR
    goal["Goal and business records"] --> task["Task"]
    task --> employee["AI employee"]
    knowledge["Knowledge and skills"] --> employee
    connector["Authorized tools and connectors"] --> employee
    employee --> result["Progress and deliverables"]
    result --> review["Your review"]
    review --> records["Linked business records"]
    records --> dashboard["Business dashboard"]

For example, in security operations (SOC):

1. Describe an alert-tracking business and review its record fields and pages.
2. Add local records, import a supported table, or configure an authorized data connection.
3. Give an analyst employee relevant procedures, evidence and tools, then assign an alert investigation.
4. Follow the task, inspect the evidence and review the deliverable linked to that alert.
5. Use a dashboard to summarize alert status, severity and trends.

Adding a solution does not automatically connect accounts, authorize employees or start scheduled work. Those choices stay with you.

Dashboard using synthetic alert records; this is not live security monitoring.

SOC is the first end-to-end business scenario. Customer delivery, content operations, retail, software development, education, store services and other security workflows are future scenario examples, rather than completed integrations. See industry examples and the SOC walkthrough.

Bring your own model service and API key. Model usage is billed by your provider; the Free software license does not include inference credits. 0.2.0-alpha.7 is available through npm and Docker Hub. Choose native npm installation for local programs, or Docker Hub for container deployment. Source builds remain available below.

Use Node.js 24.x or 22.x from 22.19 onward, npm 11.x from 11.12.1 onward, and a running local Docker engine for the dedicated PostgreSQL database. Install the exact version into a new command directory; no global installation or source build is required.

mkdir teloa-command
cd teloa-command
npm install --ignore-scripts @teloa/cli@0.2.0-alpha.7
./node_modules/.bin/teloa up
./node_modules/.bin/teloa doctor
./node_modules/.bin/teloa status

Teloa and DSH run natively on your computer. Startup opens the Web interface; configure your model in Settings. The default Web port is 3100. For a new installation, use ./node_modules/.bin/teloa up --port 3101 if it is occupied. See the installation and stop instructions, and backup and upgrades before updating an existing installation.

Install Docker with Compose and start the engine. The published image and the configuration below have been verified on Linux ARM64; amd64 has not been verified. Teloa, DSH and PostgreSQL run in containers, so host Node.js and pnpm are not required.

Download the versioned configuration into a new directory. It uses teloa/teloa:0.2.0-alpha.7 for both initialization and the application, with persistent data volumes and a generated database password.

mkdir teloa-free
cd teloa-free
curl --fail --location https://docs.teloa.ai/downloads/0.2.0-alpha.7/compose.yaml --output compose.yaml
docker compose pull
docker compose up -d --no-build
docker compose ps
docker compose logs --tail=50 app

Wait for db and app to be healthy. Successful exit of the one-time init service is expected. Open the complete authentication link from the app log, using the configured local port (3100 by default). If the port is occupied, set TELOA_PORT=3101 in this directory's .env before startup. Keep authentication links private. See installation and lifecycle instructions and backup and upgrades.

Container tools run inside the container; mounting a folder does not give control of your host applications.

Install Docker with Compose and start the engine. This route runs Teloa and PostgreSQL in containers; host Node.js and pnpm are not required.

git clone https://github.com/teloa-ai/teloa.git
cd teloa
docker compose up -d --build
docker compose ps
docker compose logs --tail=50 app

Wait for db and app to be healthy. init exits after initialization; that is expected. Open the complete authentication link printed in the app log. The default address uses 127.0.0.1:3100. If that port is occupied, put TELOA_PORT=3101 in a local .env before starting and use that port in the authentication link.

Database credentials are generated during initialization. Data persists in named volumes. To pause and resume:

docker compose stop
docker compose start

Container tools run inside the container. Mounting a folder does not give access to applications installed on your host. Use native execution when your work needs local programs.

Use Node.js 24.x or 22.x from 22.19 onward, pnpm 11.7.0, and a running local Docker engine for PostgreSQL. Teloa and DSH run on your computer, while the database runs in Docker.

git clone https://github.com/teloa-ai/teloa.git
cd teloa
pnpm install --frozen-lockfile
pnpm build
pnpm check:dsh
pnpm setup:database
pnpm setup:dsh
pnpm dev:dsh

Open the complete authentication URL printed at startup. On macOS or Linux, use TELOA_DSH_PORT=3101 pnpm dev:dsh to choose another Web port. Source-installation state stays under .runtime/; the default working directory is .runtime/teloa/workspace.

See the installation guide and local development guide for deployment and troubleshooting. Back up your data before upgrades.

1. Open Settings → Models, configure a supported provider, and verify a simple conversation. Model credentials belong in Settings.
2. Create an AI employee. Define its responsibility, select reference material, and set its tool and approval boundaries.
3. Open its conversation and make a specific request, such as: “Summarize these files, cite the sources, and list what is still uncertain.”
4. For work that needs tracking, create or approve a task. Follow actual progress, handle requests in Needs your attention, and review the deliverable.
5. Organize related work in a business or project, then add a dashboard or schedule as needed.

Guides: first task · employees · groups · businesses and dashboards · market.

Business services, runtime adaptation and the UI are organized separately:

Before submitting a change:

pnpm build
pnpm check:client
pnpm check:dsh
NODE_OPTIONS=--max-old-space-size=4096 pnpm typecheck
pnpm test:repo
pnpm test:contract
pnpm test:bindings

Run relevant backend or runtime tests for behavior changes. These may require Docker and isolated databases. Keep runtime data and credentials out of commits. See CONTRIBUTING.md for the contribution process and DCO requirements.

Application data is stored on the machine running Teloa. Remote models and authorized external tools receive the task context they need. Self-hosting does not make every model call offline. See SECURITY.md for the permission model and known limitations.

The current edition is a personal Web Alpha. Hosted multi-user services, Pro desktop/mobile clients and enterprise deployment features are outside this Free repository. External integrations require compatible services and credentials. Stopping a conversation does not guarantee immediate cancellation of every external request.

• Bugs and proposals: GitHub Issues.
• Contributions: guide and DCO.
• Product support: support@teloa.ai.
• Business inquiries: hi@teloa.ai.
• Serious security issues: security@teloa.ai, following SECURITY.md. General reports can use Issues; keep sensitive details private.

Teloa Free is licensed under Apache-2.0. Third-party components retain their own licenses; see NOTICE and THIRD_PARTY_NOTICES.md. The name and branding are covered by the trademark policy.