AI Legacy Navigator Outreach

In## Inspiration

Organizations often know that someone needs follow-up, but the process between identifying that need and actually reconnecting with the person can break down. AI Legacy Navigator Outreach was created to explore a safer, more structured approach to outreach—one that keeps consent, authorization, safety, and human oversight at the center.

What it does

AI Legacy Navigator Outreach is an AI-assisted outreach software prototype designed to help organizations identify follow-up needs, verify consent and authorization, apply safety controls, coordinate approved outreach, and record outcomes for appropriate human follow-up.

The workflow is designed around a simple principle: technology can help organizations stay connected with people without removing human responsibility from the process.

How we built it

Navigator Outreach was developed with a React frontend and TypeScript server logic using CALL-E's official SDK/API integration. The corrected canonical build includes a guarded outbound adapter together with consent gating, authorization controls, a one-call ceiling, fail-closed defaults, and permanent Do-Not-Call protections.

The corrected canonical build passed 158/158 mock/no-call tests, TypeScript verification, and a production build using Vite's runner config loader.

A separate public-safe companion was contributed to the CALL-E awesome-phone-call-agents repository. It demonstrates the CALL-E SDK import and the project's fail-closed safety pattern without exposing private code, credentials, phone numbers, or live call functionality.

CALL-E integration

Navigator Outreach integrates CALL-E through the official SDK. The outbound adapter is designed to submit one authorized follow-up request only after required consent, authorization, and safety checks pass.

During controlled testing, authorized outbound attempts did not complete a live call. The issue was reported to CALL-E Support. Therefore, the 158/158 test result verifies the integration and safety behavior under mock/no-call conditions; it does not establish successful live outbound CALL-E execution.

Challenges we ran into

The primary challenge was completing live end-to-end outbound CALL-E validation. Rather than bypassing safeguards or overstating the results, the project preserves its fail-closed controls and clearly distinguishes verified mock/no-call behavior from pending live execution.

We also created a sanitized Judge Access Build so the workflow can be reviewed without exposing credentials, private data, or controlled calling functionality.

Accomplishments that we're proud of

We completed a working outreach workflow centered on consent, authorization, safety, and human oversight. The corrected canonical build passed 158/158 mock/no-call tests, TypeScript verification, and its production build.

We also completed a public-safe CALL-E contribution, a judge-access walkthrough, documentation of the evidence boundaries, and an upstream pull request to the official CALL-E repository.

What we learned

Building responsible AI outreach is not simply about whether an automated system can make contact. The system also needs to know when not to act. Consent, authorization, fail-closed behavior, transparent evidence, and human oversight became central parts of Navigator Outreach.

What's next for AI Legacy Navigator Outreach

The next development stage is production hardening and validated end-to-end communication. The broader vision is a channel-flexible outreach platform that can coordinate approved email, text, voice, and human follow-up while preserving consent and safety controls spiration

Built With

• call-e-sdk
• node.js
• react
• typescript
• vite