Gumloop Inc., the developer of a platform that enables nontechnical workers to create artificial intelligence agents, has raised $50 million in funding.

The startup announced the Series B round on Thursday. It stated that Benchmark led the deal with participation from Shopify Ventures, Y Combinator and several others. The investment brings the company’s total outside funding to $70 million.

Gumloop’s platform is accessible through a drag-and-drop interface. Users can create an agent by specifying the AI model that should power it and entering a natural language description of the tasks it should perform. From there, Gumloop can connect the agent to any third-party applications that it may require to carry out its work. The platform offers a library of several hundred prepackaged connectors.

Users can expand an agent’s capabilities by creating plug-ins dubbed Skills. For example, a worker could equip a customer support agent with a Skill that ranks tickets based on their severity. A data visualization agent, meanwhile, can be extended with dashboard templates and a library of SQL queries.

Gumloop provides a visual editor that makes it possible to link together multiple agents into a single automation workflow. A sales team could create a workflow that uses one agent to collect data about potential buyers, another to clean the data and a third to visualize it. Workflows can also incorporate other elements such as third-party cloud services.

Customers can use their agents together with a Gumloop-developed Chrome extension. It provides the ability to upload data from an open browser tab to an automation workflow for processing. Additionally, users can configure agents to automatically repeat a certain webpage interaction at set time intervals. 

Administrators can manage their company’s agents using a tool called Gumstack. It provides the ability to define cybersecurity rules for the external tools that agents use. For example, access to an inventory database could be limited to the procurement department’s automation workflows.

Gumstack monitors what data agents access while carrying out tasks. Administrators can secure particularly sensitive data, such as the credentials with which an agent logs into a tool, using a secrets management feature. The capability is designed to speed routine tasks as refreshing credentials.

“We were in no rush to raise but the rapid, drastic improvement in LLMs over the past 6 months definitely accelerated our roadmap,” Gumloop co-founder and Chief Executive Officer Max Brodeur-Urbas wrote in a blog post. “We’ve been building the authentication-layer, enterprise observability and security features, alongside our UX for two years and the models are finally good enough to tie it all together.”

Gumloop will use the proceeds from the funding round to hire more people.

Image: Gumloop