Ali, an engineer at OpenAI, introduces “dots,” persistent cloud agents with their own identities that integrate with tools like Slack to automate and streamline workflows, such as triaging on-call incidents. By connecting dots to services like Slack and Codex, users can delegate tasks, monitor channels, and seamlessly continue investigations across devices, enhancing productivity and incident management efficiency.
Ali, an engineer at OpenAI, introduces “dots,” a persistent cloud agent designed to integrate seamlessly with existing workflows by having its own cloud computer and identity across various services like Slack. This unique setup allows dots to interact with both cloud and local computers, providing a versatile tool for engineers to automate and streamline tasks.
Ali demonstrates how he connects his dot to Slack through his Codex app, giving the dot a distinct identity within Slack. This integration enables users to tag the dot in threads to delegate tasks or automate workflows. One practical example Ali shares is setting up his dot to monitor a Slack channel during his on-call shift, where it automatically triages incoming pages and creates threads to investigate issues.
During an actual on-call incident at 3 a.m., Ali receives a page on his phone and acknowledges it. Instead of immediately diving into troubleshooting, he checks the ChatGPT app where his dot has already investigated the problem, identified the cause, and started a Codex thread for further discussion. This automation allows Ali to manage incidents efficiently without needing to get out of bed or use his laptop initially.
The Codex thread created by the dot is synchronized across devices, meaning Ali can continue the investigation seamlessly on his laptop if the issue requires deeper analysis. This cross-device continuity enhances productivity and reduces the friction typically associated with incident management during inconvenient hours.
Ali highlights that this example is just one of many potential use cases for dots. Users can customize automations to handle various tasks such as triaging feedback channels or creating pull requests directly from Slack. He expresses enthusiasm for users to explore and leverage dots to improve their workflows and productivity.