This startup thinks e-mail might be the important thing to usable AI brokers

AI firms are pushing brokers as the subsequent Nice Office Disruptor, however specialists say they’re nonetheless not prepared for prime time. AI brokers typically battle to make selections by themselves, hallucinate regularly, can’t cooperate with different brokers, fail at confidentiality consciousness, and combine poorly with present programs.

Trade pioneers like Andrej Karpathy and Ali Ghodsi have mentioned that, just like the deployment of autonomous autos, people must be within the loop to ensure that brokers to succeed. 

A startup referred to as Mixus needs to deal with that with its AI agent platform that not solely retains people within the workflow, it additionally lets customers work together with brokers immediately from their e-mail or Slack. 

“We’re assembly prospects the place they’re as we speak,” Elliot Katz, Mixus’ co-founder, instructed TechCrunch. “The place is each particular person within the workforce as we speak? For essentially the most half, they’re on e-mail. And so as a result of we are able to do that by means of e-mail, we imagine that’s a approach we are able to democratize entry [to agents].”

If Mixus works reliably, it might resolve a giant drawback within the AI agent house. Most AI firms as we speak both offer you a pre-built assistant, à la ChatGPT or Gemini, or builders need to construct customized brokers utilizing frameworks like LangChain, AutoGen, or CrewAI. 

Mixus launched in beta out of Stanford solely in late 2024, nevertheless it has already raised $2.6 million in pre-seed funding and introduced on some prospects, together with clothes retailer chain Rainbow Outlets, in addition to others throughout finance and tech. 

The startup says its largest promoting level is ease of use, from the way it helps you create brokers to how one can work together with them. Customers can use textual content prompts to arrange their brokers inside Mixus’ platform by way of a chat operate, or by merely emailing directions to agent@mixus.com. Then Mixus will construct, run, and handle single- or multi-step brokers immediately from the inbox. 

For instance, a buyer assist supervisor might use a immediate that reads so: 

Create an agent that finds all open duties in Jira in challenge mixus-dummy, and ship me a report with info on all duties which can be overdue. Draft emails to all of the assignees who’ve overdue duties, and have me overview them within the chat and with easy clear formatting for e-mail (no attachments/docs). As soon as I confirm, ship the emails. Run it now. And shifting ahead, run it each Monday at 7am PST.

Katz and his co-founder Shai Magzimof demoed the brokers for TechCrunch, exhibiting methods to add human verifiers in your brokers by merely instructing at which step the agent ought to ask you for oversight.

For instance, they ran an agent to do analysis on TechCrunch reporters earlier than pitching them. The agent recognized and gathered expertise information and developments, analyzed the data to determine potential story angles, and compiled a analysis report summarizing the findings. On the final stage, the agent was directed to ship the data to Katz for verification. As soon as accepted, the agent would ship the finished analysis report back to Magzimof. 

The founders famous that people will be within the loop as a lot or as little as required — Magzimof mentioned organizations can arrange company-wide guidelines, like making certain an e-mail will get checked by a human if it’s being despatched exterior the corporate.

Bringing different colleagues into the workflow is as simple as tagging them within the chat with an AI agent, or copying them on the e-mail to the agent. That’s one other standout in comparison with brokers on the markets as we speak: Most fashions are single-user, and whereas Notion AI and Slack permit customers to collaborate in shared areas, they don’t let the AI handle conversations and duties between teammates in actual time. 

One other core function of Mixus is its capacity to recollect recordsdata, chats, prompts, and brokers. 

“We created Areas so that each staff, each particular person, each group of individuals can have a shared reminiscence,” Magzimof mentioned. “Then all my brokers, all my recordsdata, all of the individuals will be in that very particular House’s reminiscence.”

Whereas ChatGPT and Claude each assist reminiscence, their enterprise plans don’t but assist shared agent reminiscence throughout customers.  

What else can Mixus do?

In our interview, the founders ran by means of an hour-long demo exhibiting a spread of use instances and skills. Mixus’ brokers do appear succesful, reflecting a excessive diploma of autonomy and reminiscence that locations the corporate towards the upper finish of the AI agent spectrum. That’s, if the product works as reliably because it did within the demo.

Like different brokers, Mixus can combine with different instruments, from Gmail to Jira, and customers can set off brokers to run instantly or on a schedule. Brokers can run and edit paperwork or spreadsheets inline — just like ChatGPT, Microsoft Copilot, and Google Gemini, however these are sometimes restricted to sandboxed environments.

Mixus additionally lets brokers autonomously navigate organizational context — like determining who in a corporation owns a process by trying by means of Jira tickets.

Constructed on a mixture of Anthropic’s Claude 4 and OpenAI’s o3, Mixus brokers even have entry to the net, which Magzimof says will be tapped for duties like stay analysis or monitoring. He described it as “Google Alerts on steroids.” 

Taken collectively, Mixus seems to be much less of a productiveness device and extra like a tireless digital colleague — one more bold try to reimagine AI as a collaborator. If it really works as marketed, your subsequent “coworker” won’t be human, nevertheless it may get by means of your inbox quicker than you do.

Correction: This text was up to date to mirror how a lot Mixus has raised thus far.

Bought a delicate tip or confidential paperwork? We’re reporting on the interior workings of the AI trade — from the businesses shaping its future to the individuals impacted by their selections. Attain out to Rebecca Bellan at rebecca.bellan@techcrunch.com and Maxwell Zeff at maxwell.zeff@techcrunch.com. For safe communication, you may contact us by way of Sign at @rebeccabellan.491 and @mzeff.88.