Email as the Next Big Interface for AI Agents?

AI agents are often pitched as the future of the workplace, but many businesses find them far from perfect. AI agents frequently struggle with making reliable autonomous decisions, collaborating with others, and seamlessly fitting into established workflows. Recent industry evaluations highlight these ongoing challenges. In practice, both experts and industry leaders such as Andrej Karpathy and Ali Ghodsi recommend keeping humans involved in the loop to ensure successful outcomes and build trust in agent-based workflows.

Mixus: A Human-Centric Approach to AI Agents

One startup, Mixus, is bringing a fresh approach by embedding human interaction directly into the agent experience. Rather than asking users to abandon familiar tools, Mixus’s platform enables people to interact with smart agents through their usual email or Slack environments. According to co-founder Elliot Katz, "We’re meeting customers where they are today... for the most part, they’re on email. And so because we can do this through email, we believe that’s a way we can democratize access [to agents]."

This vision resonates especially well with users unfamiliar with technical agent platforms or automation tools. Instead of building custom AI workflows from scratch, users simply send plain-language commands to Mixus. Whether the instruction is as complex as automating Jira task management or as straightforward as writing follow-up emails, Mixus seeks to keep things easy and accessible from the inbox.

Beta Launch and Early Traction

Launched in late 2024 out of Stanford, Mixus has already raised $2.3 million in pre-seed funding and brought on early adopters like Rainbow Shops, as well as companies in finance and technology sectors. The focus: usability and accessibility from day one.

How It Works: Real-World Examples

Setting up an AI agent with Mixus is as easy as sending an email to [email protected] or chatting via the Mixus platform. For example, a sales manager could request:

• Finding all overdue Jira tasks in a project
• Drafting reminder emails to responsible team members
• Submitting drafts for review before sending
• Scheduling the process to run weekly

This natural-language workflow removes barriers commonly found in more developer-centric agent frameworks like LangChain and crewAI.

Image Credits: Mixus

Collaboration and Memory: Built for Teams

Unlike most single-user agent tools, Mixus lets users bring colleagues into the loop by tagging them in chats, copying them on emails, or establishing sophisticated human checkpoints throughout an automated workflow. Organizations can create company-wide rules so that certain actions always get a human review, but they're not forced to mandate oversight if they trust the agent’s track record.

Team memory is another standout. Through "Spaces," Mixus enables persistent, shared memory between team members and agents, supporting group collaboration across all tasks and documents—beyond what is currently offered by enterprise agent products from OpenAI and Anthropic.

A running list of Mixus's AI agent capabilities. Image Credits: Mixus

Mixus vs. Other AI Agents

Aside from intuitive setup and robust collaboration, Mixus stands out for its ability to integrate with popular productivity tools like Gmail and Jira. Agents can run tasks immediately or on a set schedule, edit spreadsheets or documents inline, and autonomously navigate organizational relationships—figuring out responsibility and context without additional training.

Built on leading models such as Anthropic's Claude 4 and OpenAI's o3, Mixus agents even have live web access for research and monitoring. This positions Mixus less as just a productivity enhancement, and more as a digital colleague capable of continuous, reliable support.

Looking Ahead

It's still early days, but if Mixus’s demo reliability holds up at scale, it could redefine how businesses of all sizes use AI in their daily operations—not by replacing human input, but by streamlining access and automating routine communication through the world’s most familiar platform: email.

References

• AI companies are pushing agents as the next Great Workplace Disruptor, but experts say they're still not ready for prime time.
• They often struggle with autonomous decision-making, can’t cooperate with other agents, fail at confidentiality awareness
• Andrej Karpathy... said that, like the deployment of autonomous vehicles, humans need to be in the loop in order for agents to succeed.
• Ali Ghodsi have said that, like the deployment of autonomous vehicles, humans need to be in the loop in order for agents to succeed.
• Startup Mixus understands that and has built an AI agent platform that not only keeps humans in the workflow, but also allows those humans to interact with agents directly from their email or Slack.
• Source: TechCrunch