Skip to content

Latest commit

 

History

194 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Tel-Agent

AI phone assistant

Connect any phone line to any AI model. Self-hosted, bring your own keys.

Call    Landline    Web chat    SMS    Email    WhatsApp    Telegram    Messenger    Instagram    Discord    Slack

Microsoft Teams    Signal    Viber    Google Chat    Mattermost    Matrix    IRC    LINE    WeChat    WeCom    QQ    DingTalk    Feishu / Lark    iMessage

Connect them over MCP — and get full control.

Works with OpenAI, Claude, Gemini, Mistral, Ollama, Perplexity, Copilot, Manus, DeepSeek, Grok, Qwen, Groq and OpenRouter

Product of the day - 1st

DSGVO & GDPR compliant EU AI Act compliant

License: AGPL v3 Status: alpha Release

tel-agent.com · maintained by Dpro GmbH, Vienna


What it is

Tel-Agent is an open-source gateway that sits between a phone line and an AI agent.

A call arrives over SIP. Tel-Agent checks the caller against your routing rules and either passes it through to a human, blocks it, or hands it to an AI agent. The agent speaks with the caller in real time, can invoke tools — transfer the call, take a message, check a calendar, call any HTTP endpoint — and every call is recorded, transcribed, and searchable.

It runs on your own hardware, on your own LAN, with your own API keys.

The same agent handles phone calls and answers on web chat, SMS, email, WhatsApp, Telegram, Messenger, Instagram, Discord and Slack — connected with your own credentials from each platform, never a shared application of ours. The official channel scope now covers 24 channels including the phone, with additional channels built on the extension contract. A channel is a route a customer uses to reach you; a system you run your own business on is an integration, and those are reached through webhooks and the HTTP tool.

What it is not

  • Not a PBX replacement. It connects to your existing PBX (3CX, Asterisk, FreePBX) as an extension.
  • Not a workflow automation platform. Webhooks and a generic HTTP tool reach n8n and Home Assistant, which do that job better. A channel is where the conversation happens; an integration is a system the agent acts on. We own the first and reach the second through the HTTP tool.
  • Not a CRM.
  • Not analog-capable. Bridge an analog line with an ATA; Tel-Agent only speaks SIP.

Why it exists

Well-funded closed products already do "AI receptionist for business". What does not exist is a good open, self-hosted one — where you own the recordings, choose the models, and decide which callers ever reach the AI at all.


Project status

Alpha. Native installers and container images are available today. Choose a platform on the download page, or get the packages from the releases page: a Windows installer, macOS packages for Apple Silicon and Intel, DEB and RPM packages for Linux, and a docker-compose.release.yml that pulls the published images. The installers are unsigned while code signing is not yet configured, so the operating system will ask once before it runs.

What works now. A conversation is answered end to end: the reply streams token by token, can be interrupted mid-sentence, is stored, and raises a notification so a person knows someone wrote in. The phone implementation includes streaming speech recognition and synthesis, interruption handling, caller-ID routing, and stored call transcripts through the SIP/LiveKit transport. Twilio integration handles SMS with customer-owned credentials. The core messaging channels are wired — web chat, WhatsApp, Telegram, Messenger, Instagram, Discord, Slack, email and SMS — and the dashboard serves them from a real API: sign-in and the account flows, home, the conversation archive, the notification tray, contacts, assistants, knowledge, the catalogue, apps, numbers, routing rules, backups, system health, settings and workspaces. The product speaks five languages: English, German, Arabic, Spanish and Dutch.

What is being added now. A second wave of channels, one issue each under the channel label: Microsoft Teams, Signal, Viber, Google Chat, Mattermost, Matrix, IRC, LINE, WeChat, WeCom, QQ, DingTalk, Feishu / Lark and iMessage. They share one declarative setup contract (#220), so each new channel is a definition plus a transport rather than a fork of the last one. More translations are open as good first issues.

Deployment. The Windows installer starts the local service and opens the dashboard in your browser. Native macOS and Linux packages and Docker images provide the other installation paths. Phone and messaging connections use your own provider credentials; configure the channels you want to use in your installation.

The full plan, and what each milestone means, is in docs/ROADMAP.md. Watch or star the repository to hear when a release lands.


Architecture

Layer Choice
Voice agent Python + LiveKit Agents
API Python + FastAPI
Frontend Next.js + React
Database SQLite by default; PostgreSQL is also supported
Cache / queue Redis
Reverse proxy Caddy — automatic HTTPS
Packaging Windows installer, macOS packages, Linux DEB/RPM and Docker Compose

Providers for v1: Deepgram (STT) · one cloud LLM · ElevenLabs (TTS). Local models (Ollama, Whisper, Piper) follow in v1.1 — they need a GPU to hold a natural conversation, so they are not the default.

Latency target: under 800 ms from the end of caller speech to the first audio out. Everything streams; the first sentence starts speaking while the rest is still being generated.


Quick start

git clone https://github.com/Dpro-at/Tel-Agent.git
cd Tel-Agent
cp .env.example .env
# set ENCRYPTION_KEY in .env - generate one with: openssl rand -hex 32
docker compose up -d --build

Then open http://localhost:38471. The first visit creates the administrator — there are no default credentials. The API and its documentation are on http://localhost:38472/docs, and conversations live on the tel-agent-data volume (SQLite by default; a postgres profile is in docker-compose.yml).

Every tagged release also publishes the images to GitHub Container Registry, so the build step can be skipped entirely: docker compose -f docker-compose.release.yml up -d — same layout, same volumes, interchangeable with the from-source file on one machine. If you would rather not run Docker at all, the releases page has native installers for Windows, macOS and Linux.

Both ports are published on loopback only. Reaching the installation from other machines is a decision made in .env — the TEL_AGENT_* block there lists the three values to change and why the dashboard image is rebuilt for it. On a server, put a reverse proxy terminating TLS in front instead.

Running it without Docker stays supported and documented — contributors need to run the code without rebuilding an image on every edit. See CONTRIBUTING.md for the manual run, docs/SPEC.md for the full design and CLAUDE.md for the development rules.

Requirements for the phone channel

  • API keys for an STT, an LLM, and a TTS provider (or a GPU for local models)
  • A machine on the same LAN as the PBX
  • Network access to the PBX on 5060/UDP and an open RTP port range
  • A SIP endpoint. Most lines already are one:
Your line What you need Extra hardware
A PBX — 3CX, Asterisk, FreePBX An extension on it None
A landline from an ISP These are IP-based now. Either the provider gives you SIP credentials, or your router acts as a SIP registrar and you register against it — common with Fritz!Box in Austria and Germany None
A genuinely analog line — old copper, a fax line An ATA to bridge it, e.g. a Grandstream HT801 ~€30

Tel-Agent only ever speaks SIP. That is deliberate: supporting telephony hardware directly is a project of its own, and an ATA solves it for about the price of a cable.


Security notes

  • The port is not exposed by default. The server listens on 127.0.0.1, and python -m api warns if you change that. The supported ways to reach an installation from elsewhere — a private network, a VPN, or a reverse proxy terminating TLS — all talk to a server on loopback, so none of them requires widening it.
  • A password is required on first run. There are no default credentials.
  • API keys are encrypted at rest and are never returned in full to the client.
  • Recording announcements are on by default. Austria requires both parties to be aware that a call is recorded, and the requirement still applies once a human takes over from the agent.

Contributing

New here? docs/ONBOARDING.md is the hour-long path from a fork to a merged pull request — a translation, one file, no setup and no API keys. Thirty languages are sitting at 0% and any language you actually speak is welcome.

Contributions are welcome. Check the open issues and the channel extension contract before starting a feature, and keep each pull request focused on one concern.

All contributors must sign the CLA before their first pull request is merged. Read CONTRIBUTING.md before opening one. Everything in the repository — code, comments, commit messages, documentation — is written in English.

Built with Antigravity and Codex

Working with an AI coding agent? Point it at AGENTS.md — Codex, Antigravity and the rest find it on their own. CLAUDE.md is the full working contract; every other entry file in the repository is a pointer to those two, so the rules cannot drift apart.

Found a security problem? Do not open a public issue — see SECURITY.md.


License

AGPL-3.0. Copyright © Dpro GmbH.

If you run a modified version of Tel-Agent as a network service, you must publish your modifications. The free version is never crippled; it is the product.

For a commercial license permitting closed-source integration, contact Dpro GmbH at info@dpro.at.

Dpro GmbH · Wipplingerstraße 20/18, 1010 Wien, Austria · FN 631492s, Handelsgericht Wien

About

AI phone assistant | open-source

Topics

Resources

Contributing

Security policy

Stars

266 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages