Add Enterprise Meeting-to-Action Voice Agent

Summary

Add a voice-enabled enterprise workflow that turns meeting audio or transcripts into structured action items, risks, blockers, and follow-ups.

Proposed location

voice_agents/meeting_to_action_agent/

What to build

Create a runnable voice or transcript-driven agent that processes meeting input and extracts useful operational outputs. The agent should identify action items, owners, deadlines when present, risks, blockers, open questions, and follow-up summaries.

It should avoid guessing when ownership or dates are unclear, marking uncertainty instead. Use LiveKit, Pipecat, Whisper/STT, or another suitable voice stack, and follow the repo's existing model-provider configuration pattern.

Acceptance criteria

• Includes a README.md based on .github/README_TEMPLATE.md.
• Accepts meeting audio or transcript input.
• Extracts action items, owners, risks, blockers, and follow-ups.
• Marks unclear owners or deadlines as uncertain instead of inventing details.
• Exports structured JSON or Markdown output.
• Includes a sample meeting transcript or audio fixture.
• Adds one catalog entry to the root README.md under Voice Agents.

Contribution notes

Please keep this to one self-contained project and one pull request. Link the pull request with Closes #<issue-number>.