Hail

Universal communication platform for AI agents.
Phone calls, SMS, email — outbound first, inbound next. Self-hostable. Open source (AGPLv3).

Your agent wants to place a call: "Call +1… and ask if they want to reschedule." Hail does the carrier glue, runs the voice pipeline, and lets the agent plug in its own brain (or fall back to OpenAI → Gemini → Claude).

Quickstart

git clone https://github.com/hail-hq/hail
cd hail
cp .env.example .env
# fill in Twilio, LiveKit Cloud, Deepgram, Cartesia, and one of OpenAI / Gemini / Anthropic
docker compose up

Authenticate:

• Hail Cloud (managed at hail.so): hail login runs the device-flow and saves an API key to ~/.hail/credentials.json.
• Self-host: seed an API key directly into your local stack — see docs/operations.md "First-run DB seed". Then export HAIL_API_KEY (or pass --api-key).

Use it — CLI (for humans scripting Hail):

hail login                        # authenticate (device flow)
hail auth logout                  # remove local credentials
hail auth token                   # print bare API key for scripting

hail call +14155550100 --prompt "be brief"
hail call list
hail call tail <id>               # follow events for one call

hail email send --to [email protected] --subject hi --body "hello"
hail email list
hail email get <id>
hail email tail <id>              # follow events for one email
hail email raw <id>               # RFC 5322 source
hail email attachment <id> <att-id> --output file.pdf
hail email domain register --kind hail_mail
hail email domain register --kind custom --domain acme.com  # send + receive on your own domain
hail email domain list

hail tail                         # cross-channel event stream
hail tail call:<id>               # narrow by resource type

hail mcp endpoint                 # Streamable HTTP URL for the MCP server
hail completion zsh               # source <(hail completion zsh)
hail version

Or over HTTP / MCP:

# HTTP
curl -X POST http://localhost:8080/calls \
  -H "Authorization: Bearer $HAIL_API_KEY" \
  -d '{"to":"+15551234567","system_prompt":"..."}'

curl -X POST http://localhost:8080/emails \
  -H "Authorization: Bearer $HAIL_API_KEY" \
  -d '{"to":["[email protected]"],"subject":"hi","body_text":"hello"}'

# MCP (for AI agents — Claude.ai, ChatGPT, Claude Code, Cursor, …)
# Add a remote MCP connector in your client pointing at:
#   http://<your-host>:8081    (self-hosted)
#   https://mcp.hail.so        (Hail Cloud, later)

hail tail in action:

Full setup: docs/setup/twilio.md, docs/setup/livekit-cloud.md, docs/setup/aws-ses.md, docs/setup/mcp.md.

Tenets

1. Clear comms. Explicit OpenAPI contracts. No magic.
2. Simple code. Boring is best. No abstractions without two uses.
3. Brief docs. One screen per page. Setup ≤ 10 minutes from a fresh clone.
4. Self-hostable. docker compose up runs everything except LiveKit Cloud.
5. Pluggable brain. BYO endpoint compatible with OpenAI's completions API, or use Hail's bundled fallback (OpenAI → Gemini → Anthropic). Voice pipeline + transport are always Hail's.
6. Agent-first docs. AI agents are first-class readers. Lead with concrete, runnable examples; link to canonical sources (OpenAPI spec, MCP tool schemas, code paths) rather than paraphrase them. Every page should let a reader — human or agent — take the next action.

Milestones

Checked = shipped. Per-artifact changelogs (GitHub Releases for the CLI, PyPI release notes for the SDK) own the "shipped in which version" question.

Phone calls

• Outbound
  • Twilio
  • Telnyx
• Inbound
  • Twilio

SMS

• Outbound
  • Twilio
• Inbound
  • Twilio

Email

• Outbound
  • AWS SES
  • Custom sender domains (own DNS, auto DKIM + MAIL FROM)
• Inbound
  • AWS SES
  • Custom domains (receive on verified domains)

Voice pipeline

• STT
  • Deepgram
  • Whisper
  • AssemblyAI
• TTS
  • Cartesia
  • ElevenLabs
  • Deepgram Aura
• VAD
  • Silero
• Turn detection
  • LiveKit turn-detector
• LLM — system-prompt mode
  • Fallback: OpenAI → Gemini → Anthropic, fast models
• LLM — BYO-endpoint mode
  • OpenAI chat-completions-compatible
• Recording
  • S3 upload
  • Diarization

Distribution

• API
  • OpenAPI spec
• CLI
  • hail binary via GitHub Releases
• MCP server
  • Remote Streamable HTTP endpoint bundled with every Hail deploy
  • PyPI stdio package — intentionally not shipped; see docs/setup/mcp.md
• Python SDK
  • hail-sdk on PyPI, imports as hail

Infrastructure

• Docker Compose scaffold
• Self-hosted LiveKit SFU
  • docker compose integration

Architecture

AI agent ──► Hail API ──dispatch──► Voicebot ──► LiveKit Cloud ──SIP──► Twilio ──► 📞

Full diagram: docs/architecture.md.

Contributing

See docs/contributing.md. TL;DR: fork, branch, conventional-commit, PR. Provider adapters go in core/hailhq/core/providers/. Update .env.example for any new env var.

License

Source code: AGPL-3.0-or-later — if you run a modified Hail as a service, you must release your source.
Pricing dataset (costs/): CC-BY-4.0 — reuse the pricing JSON freely with attribution.