Here is a problem that sounds like a real-estate problem but is actually a forecasting problem. A company signs an office lease. The standard term is something like seven years. To sign it, the company has to answer a question it cannot possibly answer: how many people will work here in 2031? Guess too small and you are subletting a cramped floor in eighteen months. Guess too large and you are heating empty desks and explaining the line item to your board. Either way you have made an expensive bet on a number you invented.
Codi, founded in San Francisco in 2018 by Christelle Rohaut and Dave Schuman, is built around the idea that companies should simply not have to make that bet. Its core product is a private, fully serviced office you can rent on terms as short as six months. Rent, furniture, internet, utilities, coffee, cleaning, and a support team arrive bundled into one recurring payment. You get keys; you do not get a facilities department.
That bundling is the tell. Traditional commercial real estate is a pile of separate contracts - the landlord, the furniture vendor, the ISP, the cleaning crew, the coffee guy. Codi's move is to take that pile and turn it into a single invoice with a single number on it. This is, in the grand tradition of software companies, an argument that the interesting innovation is sometimes the billing.
Not the Airbnb it gets called
Reporters like to call Codi "the Airbnb of commercial real estate," which is a headline more than a business. The mechanism underneath is a two-sided marketplace: Codi matches companies to properties that fit their budget, neighborhood, and appetite for flexibility, then manages the move-in from design to IT to snacks - typically inside about four weeks. Landlords get vetted, shorter-term tenants; companies get to skip the buildout. a16z led the Series A, and the general partner who led it, Jeff Jordan, previously led the firm into Airbnb, Instacart, and Pinterest. So the marketplace framing is not an accident. It is the thesis.
The contrast Codi likes to draw is with WeWork. WeWork sold you a desk on a shared floor, surrounded by strangers and kombucha. Codi's argument - stated more or less directly in its Series A messaging - is that people never actually liked that. What a lot of teams want is a private room of their own, used maybe a couple of days a week, without a decade of obligation attached. When WeWork's model faltered, the contrarian version started to look less contrarian.
The urban planner in the room
It helps to know that Rohaut is not a real-estate broker by training. She holds a master's in city and urban planning from UC Berkeley, and she talks about offices the way a planner talks about neighborhoods - as spaces that either work for the people inside them or quietly don't. That background is the most useful thing to know about Codi, because it explains a company that keeps treating "the office" as a design problem rather than a square-footage problem.
It also explains the arc. Codi's earliest incarnation, around 2020, was stranger and smaller: a $7 million seed round behind the idea of turning homes and living rooms into local, neighborhood coworking spots. That version didn't become the company. But the instinct - space should bend to how people actually live and work - survived every pivot since.
Then it built a robot office manager
The newest chapter is the most software-shaped. In October 2025, after a beta earlier that year, Codi launched what it calls the first AI platform to fully automate office management. The unglamorous truth of running an office is that it is mostly coordination: restock the pantry, chase the cleaner, find a vendor, approve the invoice. Rohaut has put a number on that drudgery - office admin, she says, can run north of $80,000 a year - and the pitch is that coordination is exactly the kind of work software is good at.
The early numbers were loud enough to notice. Codi says the AI Office Manager reached $100,000 in annual recurring revenue within five weeks of its beta, signing forty companies including TaskRabbit and Northbeam. It is sold as a monthly management fee that Codi positions as a fraction of a part-time office manager's cost - which is a very software way to price a very physical job.
Put the three acts together and Codi looks less like a company that pivoted and more like one that kept following its own logic downhill. First make the lease flexible. Then make the office turnkey. Then automate the running of it. Each step removes one more thing a company has to think about. Whether that adds up to a durable business or a very convenient one is the open question - but as unbundling-and-rebundling stories go, this is a tidy one.