Evos bags $9M to build autonomous workers that learn from companies’ most experienced humans

Evos bags $9M to build autonomous workers that learn from companies’ most experienced humans

A startup called Evos said today it has raised $9 million in pre-seed funding to spearhead the adoption of artificial intelligence-based automation within smaller businesses by learning from their most valuable assets – the people they employ.

Today’s round was led by Uncork Capital and Field Ventures and was joined by Crosslink Capital, Utopia Capital Management, Sandwith Ventures and a number of anonymous angel investors.

Evos, officially known as Evolutionary Operating System Inc., is pioneering an agentic operations platform that enables companies to create autonomous operators that can support workers across the physical economy, in industries such as manufacturing, freight and logistics. These AI agents are designed to work autonomously, but only after learning from the people they’re designed to assist.

Founder and Chief Executive Urav Shah stressed that he’s not trying to replace humans with autonomous workers, but instead help companies to solve a growing problem. Many small and midsized industrial businesses today are facing serious labor challenges, because they’re unable to replace the seasoned veterans that currently form the backbone of their business. According to Shah, in many industries these skilled workers are retiring faster than they can be replaced.

This is a big problem, because most smaller companies with less than 100 employees run their business in a very specific way, and their most valuable assets are the people who understand those processes. That knowledge lives in their heads, and as they leave the workforce, businesses need a way to pass on that knowledge. “The most capable system in many of these businesses is the person who has been there for 20 years, and what they know has never been written down,” Shah said.

Learning on the job

Evos’ proprietary agentic operations platform, called Workhorse, is designed to automate the capture of this knowledge, so that its autonomous operators can start doing real work and free up humans to spend time on higher level tasks, Shah explained. But unlike other agentic AI tools, which are simply assigned a task and told to get on with it, all Evos’ agents do at the beginning is sit and watch.

Each agent is given a specific task to master, such as answering broker calls, chasing missing paperwork, reconciling invoices or building freight loads. It sits behind an “approval gate,” which means that humans have to review, approve and modify or reject its decisions and actions. When it does the task correctly, this becomes training data for the underlying model that powers the agent, enabling it to learn on the job.

Only once the agent has fully learned its role, and proved its expertise in a real-world setting with constant human oversight, is it freed from its bonds to start doing that particular job autonomously. And if it ever comes across something new, that it hasn’t seen before, the decision it makes or action it takes will need to be approved by a human first.

“An Evos operator learns the job from these people, improves on it and proves it can match them before it takes on any work by itself,” Shah stressed. “Our goal is to put that expertise on every job, every day, so these businesses can run like the best in the world do.”

Shah admitted that Evos’ autonomous operators aren’t entirely perfect, but he claimed they’re still pretty good. The average agent can achieve roughly 82% decision accuracy after just 24 hours of learning, at which point they can handle around 70% of any human’s daily workload fully autonomously. And from there, they just get better and better, Shah said.

Field Ventures Partner Richard Kerby said the vast majority of businesses in the physical economy use software primarily to organize and track work. But there are very few tools that actually do the work. “Evos changes that,” he said. “Its autonomous operators learn the judgment behind how a business runs, work within existing systems and take ownership of business functions.”

The freight operator MyFreightWorld is using Evos’ agents to audit each less-than-truckload carrier invoice against its transportation management system. In the first two months of its deployment, it made more than 3,500 auditing decisions with 97% accuracy, helping to process a total invoice value of more than $600,000.

That’s skilled work that would have taken dozens of hours for a human to perform. “Evos has helped us revolutionize our billing process with zero upfront investment and do it at an impressively fast speed measured in weeks and days,” said MyFreightWorld President Mike Chalfant.

Evos went live with select customers only a month ago and has already reached more than $1 million in annual recurring revenue, Shah said. The company plans to make the Workhorse platform generally available later this year, and will use the money from today’s round to scale that rollout and grow its teams.

Uncork Capital’s Tripp Jones said the thing that stood out for him was how fast companies are able to put Evos’ agents to work. “It puts powerful technology within reach of businesses that software has long overlooked,” he said.

Images: Evos