People who build unconventional businesses face some things that are structurally different to those who become something legible: a butcher, a baker, a candlestick maker.
I fucking loved Lentil as Anything. Vegan, pay as you can, training and work for marginalised people in several premium city locations. Ran for twenty odd years. Total trailblazers. Legends.
I was sad when it closed, but not all that surprised. When you have an unconventional business model you’re often fighting against additional things. It’s worth it. So worth it. But I thought maybe if we looked at some businesses like Lentil, we could learn a bit about what went wrong — so that when we’re building or trading in an unconventional way we can build in resilience and hopefully have longer successes, wider communities and bigger impacts.
Before we start: a business closing is not a business failing. A business that served people and made a difference for any amount of time did not fail. Business owners often frame it that way and they’re wrong. What follows is not an autopsy. It’s an attempt to understand what made some extraordinary businesses vulnerable, so the next versions of them last longer.
Lentil as Anything
What they did: Vegetarian and vegan restaurants across Melbourne and Sydney where customers paid what they felt the meal was worth — or volunteered, or ate for free. They also employed refugees and asylum seekers and functioned as community spaces and mutual aid networks alongside being restaurants. Several locations, premium city real estate, over twenty years of operation.
What made them special: They ran for over twenty years on a model that the hospitality industry said couldn’t work. They fed people who couldn’t afford to eat out with the same food, in the same space, as people who could. They gave work to people who couldn’t easily get it elsewhere. They built something that felt, to everyone who walked through the door, like proof that a different kind of business was possible. Melbourne loved them. A lot of us still do.
What everyone said when they closed: People didn’t pay enough. The model was too idealistic. You can’t run a restaurant on trust.
What else was in the picture: The restaurant industry runs on margins so thin that a bad month can end a business that’s been trading for years. Lentil removed the one mechanism restaurants use to protect those margins — fixed pricing — and replaced it with community goodwill. For twenty years, that largely worked. Then COVID hit.
The international backpacker and traveller community made up a significant portion of their volunteer base — the people who kept the kitchens running without a wage bill. When international travel stopped, that pipeline dried up almost overnight. They kept going. They provided around 100,000 free meals during lockdowns while their income collapsed.
When administrators were called in, the picture that emerged included allegations of unpaid wages and superannuation, potential insolvent trading, and financial mismanagement. The public conversation focused on customers who didn’t pay enough. The structural reality was that the model had very little protection against any significant external shock, and when the shock came it exposed governance and financial problems that had likely been accumulating for some time.
The generosity was real. The impact was real. The vulnerability was also real, and it was structural.
Panera Cares
What they did: A pay-what-you-can version of the Panera Bread café, backed by the full resources of a major US chain. Customers could pay the suggested price, pay more, pay less, or volunteer their time instead. Five locations across the US, operating from around 2010 to 2019.
What made them special: This wasn’t a charity arm or a PR exercise. It was a genuine attempt by a large corporation to build dignity-based food access into the daily operations of a commercial business. The idea was that people who could pay more would subsidise people who couldn’t, and the whole thing would roughly balance out. In concept it was elegant. In practice it was one of the more honest experiments a major food chain has ever tried.
What everyone said when they closed: People abused it. The homeless took over. It wasn’t viable.
What else was in the picture: The cafés recovered between 60 and 85 percent of their costs on average. That’s a gap, but not necessarily a fatal one for a company the size of Panera. What actually broke the model was something harder to fix than a funding shortfall.
The spaces became genuinely confusing to everyone in them. Paying customers weren’t sure if they were in a café or a charitable service. Some locations became informal gathering places for people experiencing homelessness — not because anything had gone wrong exactly, but because the model had no mechanism for managing what the space was and wasn’t. Staff found themselves navigating complex social situations they had no training or support for. Security became an issue in some locations. The experience for the people the model was designed to help — people who needed a dignified meal without enough money to pay for it — was often undermined by the chaos around them.
The model asked everyone involved to hold a lot of ambiguity without giving them any tools to do it. That ambiguity compounded over time until the spaces stopped functioning as either good cafés or good community resources. Even with a corporation’s resources behind it, unclear boundaries and unclear roles proved fatal.
Buy Nothing Project
What they did: Hyper-local gift economy groups where neighbours gave and received items for free. No money, no barter, no debt. You posted what you had. You asked for what you needed. Someone nearby said yes. It started in the US in 2013 and spread globally, eventually reaching millions of participants across thousands of local groups, most of them operating through Facebook.
What made them special: It worked. Remarkably, simply, it worked. People gave each other things. Neighbourhoods that had never spoken found reasons to knock on each other’s doors. Waste reduced. Relationships formed. It demonstrated something important — that people will be generous with each other when given a simple, low-friction structure to do it in. For years it was one of the purest examples of a relational model operating at genuine scale.
What everyone said when they struggled: The organisation tried to own a movement. They went corporate. They killed what made it special.
What else was in the picture: The grassroots model was genuinely thriving and continued to thrive in many places. The strain came from a specific set of decisions at the organisational level — launching a proprietary app to move communities off Facebook, enforcing trademarks, centralising governance over what had always been a decentralised and volunteer-driven network.
In late 2025, hundreds of independent Facebook groups were shut down over trademark disputes with the official organisation. Communities that had been running for years, built by volunteers, serving real neighbourhoods, were told they couldn’t use the name. Some complied and rebranded. Many dissolved. The people who had built those groups experienced it as the organisation claiming ownership over something that had never belonged to it.
The tension wasn’t between generosity and selfishness. It was a structural one: a model that worked precisely because it was decentralised, relational, and ungoverned came into conflict with an organisation that needed to centralise, formalise, and govern in order to survive as an entity. Those two things turned out to be incompatible. The movement survived in fragments. The organisation’s relationship with its own community did not.
What These Three Have in Common
The impact was genuine, the communities were genuine, the values were genuine. None of them closed because the idea was wrong or the people were naive. They ran into specific, structural vulnerabilities that unconventional models carry and conventional ones largely don’t.
The first is dependency on conditions they couldn’t control. Lentil as Anything built their operations around a volunteer base that evaporated when the world closed. The model worked beautifully in normal conditions and had almost no resilience when conditions changed. Conventional businesses carry this risk too, but they tend to have pricing mechanisms and financial buffers that create some protection. Lentil had given those away in service of the model itself.
The second is the absence of boundaries. Panera Cares had resources, corporate backing, and genuine intention. What it didn’t have was clarity about what the spaces were, who they were for, and what was expected of everyone in them. Without that clarity the spaces became something nobody had designed and nobody could manage. Boundaries aren’t the opposite of generosity. They’re what makes generosity sustainable.
The third is the tension between a relational model and organisational scale. Buy Nothing worked because it was local, human, and ungoverned. The attempt to put a formal structure around it introduced exactly the dynamics — ownership, control, centralisation — that the model had been built to refuse. Some things scale. Some things replicate instead of scale, and trying to do the wrong one breaks them.
SAME Café in Denver is still running after nearly twenty years on a similar model to Lentil as Anything. The differences are instructive: smaller scale, stronger local embedding, clearer participation expectations, hybrid funding that doesn’t depend entirely on customer payments, and slower more deliberate growth. The values are the same. The protective structures around those values are more robust.
That’s probably the thing worth taking from all of this. The businesses that last longest aren’t the ones that compromise their values. They’re the ones that build adequate protection around them. Generosity needs infrastructure. Community needs boundaries. Radical models need resilience planning, not because the world is hostile to them — though sometimes it is — but because they’re often operating without the conventional safety nets that more legible businesses take for granted.
Closing is not failure. Twenty years of feeding people who couldn’t afford to eat is not failure. A neighbourhood gift economy that brought millions of strangers to each other’s doors is not failure. But understanding what made them vulnerable is how the next version of those businesses lasts thirty years instead of twenty, reaches further, and holds together when the world gets difficult.