French startup Mud has raised a $16 million Collection A funding spherical led by Sequoia Capital. With Mud, firms can create customized AI assistants and share them with their staff in order that they’ll work extra effectively.

However what’s fascinating with Mud is the variations with different firms engaged on enterprise brokers or AI assistants generally. Not like a consumer-facing device like ChatGPT, Mud assistants are related to an organization’s knowledge and paperwork. As an example, once you construct a brand new assistant in Mud, you’ll be able to affiliate it with Notion pages, paperwork saved in Google Drive, Intercom conversations or Slack.

On the identical time, in contrast to most AI startups engaged on enterprise brokers, Mud believes that firms ought to have a number of AI assistants — not only one. Every assistant could possibly be helpful to carry out a sure set of duties and resolve some frequent issues {that a} particular group is dealing with.

In additional sensible phrases, help groups can use a Mud assistant that’s conscious of each the content material of the data base and previous help interactions. This fashion, new group members within the help group can ask a query to the @supportExpert assistant and get a related reply.

HR groups can create an AI assistant that may reply questions on company insurance policies — no want to look a convoluted Notion database. They’ll additionally create a distinct agent that may draft job descriptions primarily based on previous job descriptions. As soon as once more, this empowers the corporate at massive and frees up time for the HR group.

For engineering and knowledge groups, the use instances are fairly simple. For instance, a Mud assistant can pay attention to the corporate’s database schemas. You’ll be able to ask @SQLbuddy in plain English to jot down a SQL question in your buyer base.

One final instance: gross sales groups can generate draft emails primarily based on CRM knowledge and the overall context behind a possible consumer. And if it’s essential to create your personal connectors or combine Mud assistants in one other device, the corporate affords an API.

As an alternative of reinventing the wheel, Mud focuses on constructing a product that works for everybody. A few years after the launch of ChatGPT, most individuals are actually aware of AI assistant (many are even utilizing it for work although its in opposition to firm insurance policies). They know the way to begin a dialog, observe up with extra particulars and ask the AI assistant to reframe its reply.

Utilizing Mud isn’t that totally different as firms are constructing conversational assistants with the platform. Workers can then go to Mud’s net interface or work together with assistants in Slack straight — this manner, they are often @-mentioned in the course of a dialog. Mud basically needs to show generative AI into an inner communication device that everybody makes use of day by day.

The startup now generates $1 million in annual recurring income with some late-stage tech firms utilizing it intensively, resembling Watershed, Alan, Qonto, Pennylane and PayFit.

Enterprise banking startup Qonto estimates that 75% of its group of 1,600 are utilizing Mud assistants on a month-to-month foundation. At Alan, a French medical health insurance unicorn, 80% of the corporate makes use of AI assistants on a weekly foundation. Accounting tech unicorn Pennylane has created 86 customized assistants with Mud.

Along with Sequoia Capital, a number of the startup’s present traders are investing as soon as once more, resembling XYZ, GG1, Join Ventures, Seedcamp and Motier Ventures.

Having a customer-focused strategy additionally implies that Mud isn’t creating its personal basis mannequin. Once you construct an assistant, you’ll be able to decide the massive language mannequin that you simply wish to use for that assistant. Mud has integrations with OpenAI (GPT), Anthropic (Claude), Mistral and Google for its Gemini fashions.

There are fairly just a few startups engaged on enterprise platforms for constructing AI brokers. Some names that come to thoughts are Brevian, Tektonic AI, Ema, Kore.ai and Glean. Even Atlassian, the enterprise software program big behind Jira and Confluence, has launched its AI teammate Rovo. Let’s see if Mud has discovered the precise go-to-market technique with its straightforward onboarding technique.