In the field of artificial intelligence, an increasing number of companies are viewing AI agents as the new engine of a workplace revolution. However, experts point out that current AI agents are still immature, often making decision errors, providing incorrect information, lacking teamwork capabilities, and having poor awareness of privacy protection. To address these challenges, Mixus, a startup from Stanford University in California, is launching an innovative AI agent platform that allows users to interact directly with AI agents via email or Slack.

Elliot Katz, co-founder of Mixus, said: "We want to provide convenience in users' daily work because most people use email." A highlight of Mixus is its user-friendly design, allowing users to create and manage their own AI agents through simple text prompts or by sending emails to a designated address. This flexibility makes AI more accessible.

Internet of Things, Internet

For example, a customer support manager can create an agent by sending a simple instruction, which helps her automatically query incomplete tasks in the project management tool Jira and generate reports. After completing the task, the agent will also send an email draft for review to the manager, ensuring human involvement in the review process.

Another unique feature of Mixus is its ability to share team memory. Co-founder Shay Margalit said: "We created space functionality so that team members can share information and files, and all AI agents and files can be centralized within specific spaces." This shared memory design goes beyond the functionality of many AI tools currently on the market, enhancing collaboration efficiency.

Aside from email and Slack, Mixus can also integrate with various tools such as Gmail and Jira, allowing users to choose to run the agent immediately or trigger it on a schedule. Additionally, Mixus's agents can independently understand task assignments within the organization, finding the colleagues responsible for tasks by analyzing Jira tickets.

If Mixus operates as advertised, it could become a major breakthrough in the field of AI agents, helping users handle work tasks more efficiently and making the next "colleague" an AI assistant rather than a human.