Founder File / No. 09
Steve Johnson
He taught the early internet to show pictures. Now he is teaching machines to read the paperwork nobody wants to.
The economist who became a compression engineer.
He taught the early internet to show pictures. Now he is teaching machines to read the paperwork nobody wants to.
There is a stack of paper sitting in a healthcare intake office right now - faxed orders, handwritten charts, insurance forms - and someone is retyping it by hand. Steve Johnson has spent his career making that someone's job disappear, gently.
At Notable Systems, the Denver company he co-founded and runs as CEO, the product is unglamorous and enormous: take the cluttered documents that clog medical billing, and turn them into structured, usable data. The method is a blend - AI and machine learning doing the heavy lifting, with human reviewers in the loop where accuracy is non-negotiable. The payoff Johnson sells is concrete: faster claims, fewer denials, and hours of a person's day handed back to them.
In April 2025 the company raised a $12 million Series B, co-led by Harbert Growth Partners and Grotech Ventures, on top of an $8.8 million Series A the year before. The new money funds a deliberate expansion - Notable Systems started narrow, automating order intake for medical equipment providers, and is now reaching across the whole revenue cycle. Total funding sits at $27.6 million.
The sequencing matters. Rather than promising to automate everything at once, Johnson picked one painful, well-defined workflow - getting orders into the system correctly - and earned the right to widen out from there. Claims processing, denial prevention, coverage criteria review: each is a fresh stack of documents, and each is a place where reading faster and reading right turns directly into recovered revenue. The Series B is permission to climb the rest of that ladder.
It is the third act of a pattern Johnson has been running since 1991. Every company he has built answers the same quiet question: when there is too much information to handle by hand, how do you find the part that matters? In the 1990s the answer was compression. In the 2000s it was personalization. Today it is document AI. The wrapper changes; the instinct does not.
He did not arrive here through a straight line, and that is the point. Catch him mid-stride and you find a Harvard economics PhD candidate who walked away to write image-compression code, a man who negotiated one of the internet's defining acquisitions, and a founder who keeps choosing the unsexy problem because the unsexy problem is usually the real one.
Healthcare runs on documents, and the documents are a mess. A medical equipment provider receives an order as a fax, a scanned PDF, a handwritten note, an attached chart - rarely as clean structured data. Before anyone can bill for anything, a person has to read all of it, decide what it says, check it against insurance rules, and retype it into a system. Multiply that by every order, every day, and you get the backlog Notable Systems was built to drain.
Johnson's framing is deliberately unsentimental. The company exists to fix "the paperwork problem" - the hours intake teams spend reviewing, organizing and re-entering data instead of doing work that requires a human brain. The engine combines OCR and natural language processing to pull meaning out of cluttered pages, machine learning to classify and route, and human reviewers positioned exactly where the cost of a mistake is highest.
That last piece is a philosophy, not a hedge. Plenty of AI companies sell full automation and quietly hope the errors stay small. Notable Systems keeps people in the loop on purpose, because in healthcare a misread number is a denied claim or a delayed treatment. The promise is not a robot that replaces the intake team; it is a system that does the dull 80 percent so the team can handle the 20 percent that actually needs judgment.
It is the same trade Johnson has always made. Compression threw away the bits a human eye would never miss and kept the ones it would. Personalization threw away the recommendations you would never click and kept the ones you might. Document AI throws away the manual keystrokes and keeps the human exactly where the stakes demand one.
"The idea is that you can carry not just your credit card but your personality and preferences with you everywhere."- Steve Johnson, on the vision behind ChoiceStream
Johnson-Grace did not set out to change the internet. When Steve Johnson and Chris Grace founded it in Newport Beach in 1991, the plan was generic spreadsheet technology for simulation modeling - the kind of product that wins no headlines. Then they sat down with Irving Reed at USC, who had an idea for a better image-compression algorithm. They started building it. That detour became the ART file format.
The timing was perfect and nobody knew it yet. Bandwidth in the early 1990s was a trickle; a single photograph could take minutes to load, if it loaded at all. Johnson's compression made images small enough to actually move across a phone line. In 1993, America Online used his patented algorithm to deliver its first online pictures - the technical seed of what would become "You've Got Pictures" and, more broadly, streaming media.
AOL noticed. On February 1, 1996, it acquired Johnson-Grace for roughly 1.6 million shares. Johnson did not cash out and vanish. He became VP of AOL's Software and Technology Development division at the exact moment AOL was becoming, in the language of the era, America's front door to cyberspace. He shipped AOL 4.0. He brought telephony and broadcast products to market. He ran media-technology research.
And in 1998, when AOL acquired Netscape, Johnson was the principal technology negotiator - then led the team that folded Netscape in and helped form the Sun-AOL iPlanet alliance. His US patent for the compression technology was issued in 1999. By then the web had pictures, and most people had no idea whose math had put them there.
Make heavy media light enough to travel. The ART format and the AOL patent turned the early web from a wall of text into something you could look at.
ChoiceStream tackled the opposite problem - too many choices. It learned what you watched, rated and bought, then served media brands recommendations that fit.
Notable Systems reads the documents people dread - orders, charts, insurance forms - and outputs clean data, with humans checking the machine where it counts.
The resume reads backwards from what you would expect. Johnson took a B.A. in Economics from USC in 1980, then a Master's in Public Policy from Harvard's Kennedy School in 1985, then stayed on in Harvard's doctoral program in Economics and Philosophy until 1990. That is a decade of training to become a professor, or a policy economist, or a think-tank lifer.
Instead he founded a compression company. The leap from philosophy of economics to file formats is less strange than it looks: both are about modeling the world efficiently, throwing away what does not matter and keeping what does. Compression is information theory you can sell.
Since leaving AOL, Johnson has worked as a technology entrepreneur and investor, founding companies in automated personalization, data extraction and advertising technology, and has spent as a philanthropist with a focus on education, the arts and LGBTIQ rights. He has also served as a mentor at the venture firm Mucker Capital, advising the next batch of founders working the same edge he has worked his whole life.
Notable Systems is not a first-timers' startup. It is a reunion of operators who have shipped at scale before.
Distinguished engineer who helped architect and scale AOL's core infrastructure and AIM as it grew past 53 million users - leading hundreds of engineers through networking and reliability fires.
The team brings deep healthcare and data experience from companies like Lincare, Enovis and Numotion - the people who know exactly where the paperwork piles up.
Faster claims. Fewer denials. More time for the work that actually matters.- The Notable Systems pitch, in eight words