# Send to Dot: Bridge Phone Photos to Existing ChatGPT Conversations

A custom integration tool enabling seamless image sharing into active AI chat sessions.

- **Source:** [Hacker News: Show HN](https://news.ycombinator.com)
- **Published:** 2026-10-31
- **Topic:** [Software & Engineering](https://brifing.si/c/engineering)
- **Original article:** <https://github.com/bryanzk/send-to-dot>
- **On Brifing:** https://brifing.si/s/ddfc6634-2671-4db7-8325-4472932ddf2f

## Solving Chat Fragmentation

The Send to dot project addresses the frustration of ChatGPT creating new chat sessions whenever a user uploads a photo from their mobile device. By building a custom bridge, this tool allows users to inject images and questions directly into an ongoing conversation, maintaining context without manual re-uploading or session switching.

## Infrastructure and Requirements

Users must provide their own always-online Mac, a stable HTTPS address, and network access to host the service. The project does not provide hosting, model API keys, or author credentials. It relies on the user's existing ChatGPT account, which must support custom app connections and OAuth for secure authentication and authorization.

## Guided Installation Process

Installation is a manual, guided process rather than a one-click setup. Users download the latest release from the official GitHub repository and follow the provided INSTALL_FOR_DOT.md file. This requires configuring the service on a local Mac, setting up OAuth, and ensuring the environment is properly paired with the ChatGPT custom app.

## Operational Mechanics

The system functions as an asynchronous delivery pipeline. When a user shares a photo, it passes through the user's HTTPS service before reaching ChatGPT. The tool requires specific native tools within the dot conversation to save the original image privately and display it as an assistant attachment, ensuring a clean, integrated user experience.

## Security and Privacy

Privacy is prioritized by design, as the service only processes photos explicitly chosen by the user. It does not scan device galleries or impersonate user messages. However, users must be cautious with sensitive data, keeping passwords, owner codes, and tunnel secrets out of chat logs, repositories, and shared mobile Shortcuts.

## Performance and Limitations

Delivery is asynchronous and does not guarantee sub-30-second performance. The system currently supports single PNG or JPEG uploads up to 8 MiB, while HEIC, PDF, and video formats remain unsupported. Furthermore, automatic gallery synchronization is not included, requiring users to manually select and send each image to the service for processing.

## Authorization and Expiry

Access grants for uploads and ChatGPT connectivity typically last 30 days. These grants require manual renewal or reauthorization, as upgrading the code does not automatically extend existing sessions. Hourly event leases refresh independently, and any interruption to the Mac's power or network connectivity will immediately halt the delivery of new images.

## Project Status and Validation

While successful image-and-question sharing has been verified in real user environments on both Android and iPhone, long-term stability remains unverified. The project is licensed under MIT and serves as a technical solution for power users who require deeper control over their AI interaction workflows and existing conversation history management.

## Sources covering this story

- https://github.com/bryanzk/send-to-dot — 

---

_Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._
