Seven Years Before Slack: The Origin Story of Campfire (37signals)

Category: Company Biographies | Anti-Unicorn / Bootstrapped / Dev Tools / SaaS Philosophy
ICP Relevance: RevOps, CTOs, Eng Leaders, anyone wrestling with build-vs-buy or the subscription economy
Stacksync Angle: They built Slack seven years before Slack. They refused the money that would have let them scale it. The irony is the whole story.

THE HOOK: The Chat App That Should Have Been Slack

On February 15, 2006, a small software company in Chicago quietly launched a product called Campfire.

It was group chat for business teams. Web-based. No client to download. Files transferred reliably. Persistent logs. A permanent URL for every room. You could drop an image in and it would render inline. Your whole team could be in the same conversation from different cities, different time zones, different machines — and it just worked.

Sound familiar?

Seven years later, in August 2013, a team emerging from the ashes of a failed video game called Glitch announced a product called Slack. Searchable Log of All Conversation and Knowledge. Group chat for business teams. Web-based. Files. Images inline. Persistent history. By February 2014, it had 120,000 daily users in its first week of public release. By 2021, Salesforce bought it for $27.7 billion.

The company that built Campfire watched all of this happen.

They had built Slack. Seven years before Slack. They had the product, the users, the infrastructure, and the philosophy. What they did not have — and what they would never allow themselves to have — was the hunger to grow at any cost. And that choice, to stay small and stay free, is what makes 37signals one of the most fascinating, maddening, and instructive companies in the history of the internet.

THE BACKSTORY: Two Men Against a Valley

In 1999, a twenty-something designer named Jason Fried founded a web design consultancy in Chicago. He named it 37signals, after the 37 radio signals identified by astronomer Paul Horowitz as potential extraterrestrial communications — anomalies in a universe of noise that might mean something. It was a fittingly oblique name for a company that would spend the next two decades being an anomaly itself.

The early 37signals was a small shop. Three co-founders — Fried, Carlos Segura, and Ernest Kim — building websites for clients. Chicago, not Silicon Valley. Services, not software. They were good at it, but the real trajectory of the company changed in 2003 when Fried hired a Danish programmer named David Heinemeier Hansson to solve an internal problem.

DHH was twenty-three years old, based in Copenhagen, and brought into the company remotely to help build a web application that would help 37signals manage its own design projects. He chose to work in a then-obscure language called Ruby, and to structure his approach around a set of conventions he was developing as he went. That internal project became Basecamp. The conventions became Ruby on Rails. Neither decision was accidental, but neither was entirely planned. They were the natural products of a specific philosophy — that software should be simpler than everyone was making it, that the tools should work for the people using them, and that complexity was almost always a choice.

DHH is the ideological foil to Silicon Valley's entire operating system. Born in Copenhagen in 1979, competitive from the start, he is the kind of person who, once he decided to take up motorsport, did not dabble — he eventually raced the 24 Hours of Le Mans twelve times, won his class in 2014 driving an Aston Martin with an all-Danish team in the GTE Am category, finished second in LMP2 in 2015, and third in 2016. He brought the same architecture to software: understand the system, choose your conventions deliberately, commit entirely, and do not apologize for your pace.

Jason Fried operates differently — quieter, more measured in public, the kind of thinker who publishes essays that read like they were written in one sitting but clearly weren't. Together they form an unusual pairing: the passionate provocateur and the patient philosopher. What they share is an almost physical aversion to bullshit — in management, in software design, in the startup mythology that treats growth as a moral virtue.

THE GRIND: Building the Stack on Their Own Terms

Between 2004 and 2010, 37signals built something unusual: a portfolio of profitable web software products, all bootstrapped, all built by a small team, all generating real revenue without a dollar of outside investment.

Basecamp shipped in 2004. It was originally their own internal project management tool — built because client work required managing projects and nothing on the market did it cleanly. They launched it as a paid product almost as an afterthought, and it immediately resonated. It was simple in a way that project management software almost never was. No Gantt charts. No resource allocation matrices. Just milestones, messages, to-do lists, and files. The kind of tool that a real team actually used instead of the tools real teams were supposed to use but didn't.

DHH extracted the framework he'd built underneath Basecamp and released it separately in July 2004 as Ruby on Rails. He open-sourced it. Gave it away. The first International Ruby Conference in the US that year had 42 attendees. DHH asked how many were using Ruby commercially. He raised his own hand. One other person raised theirs.

What followed was one of the fastest adoption curves in programming history. Rails introduced two ideas that reshaped how people built for the web: Convention over Configuration (you don't spend hours setting up your environment — the framework makes sensible default choices and you override when needed) and Don't Repeat Yourself (every piece of knowledge lives in one place). Combined with Ruby's expressive, almost English-like syntax, these principles made building web applications genuinely faster. Not marginally faster. Dramatically faster.

DHH recorded a video demonstration in 2005 showing how to build a functional blog application in fifteen minutes. For developers used to Java frameworks that required XML configuration files longer than the application itself, it looked like magic. It was not magic. It was opinionated software design taken seriously.

Within two years of that video, the first RailsConf drew 300 people. The next year, 800. The year after, 2,200. Twitter was built on Rails. GitHub was built on Rails. Airbnb was built on Rails. Hulu was built on Rails. The framework that DHH extracted as a side effect of building an internal tool for a twelve-person web consultancy in Chicago powered some of the most valuable companies on the internet. Over 1.2 million web applications have since been built on Rails. The framework has been downloaded more than 130 million times.

And all of it started because 37signals needed to organize their own client projects.

Back in Chicago, the product portfolio kept expanding. Backpack launched in 2005 — a personal organizer and intranet for small teams. Then Highrise in 2007 — a lightweight CRM. The team stayed small, under fifty people for most of its history. They never raised outside capital. Every product was profitable. The blog — Signal vs. Noise — became required reading for a generation of founders and designers.

In 2006, they launched Campfire.

CAMPFIRE: The Product That Time Forgot

Campfire launched on February 15, 2006. Its premise was simple and correct: teams needed a place to talk in real time, and email was the wrong tool for that.

The original product had everything that would later make Slack compelling: web-based, no software to install, persistent message history, file uploads, inline image rendering, permanent URLs for chat rooms. It was faster than email for quick questions. Less intrusive than a phone call. Searchable. Archived. The product worked.

By any reasonable measure, Campfire was a hit. Hundreds of thousands of users adopted it. Over the years of its operation, more than a billion messages were sent through the platform. It attracted teams, agencies, startups, and remote workers who needed a lightweight real-time communication layer that wasn't email.

But 37signals never treated Campfire as a platform. They treated it as one product among several — profitable, useful, maintained, but not a mission. They did not hire armies of growth engineers. They did not chase enterprise contracts with dedicated sales teams. They did not raise a Series A to pour into virality experiments. They ran Campfire the same way they ran everything: with a small team, attention to quality, and no appetite for the kind of growth that required selling your independence.

In 2013, they quietly stopped accepting new customers for Campfire. The standalone product was wound down and its functionality folded into Basecamp 3 as a built-in feature. There was no dramatic announcement. No press release. Just a retirement page on the website.

That same year, Stewart Butterfield's team at Tiny Speck — recovering from the collapse of an online game called Glitch — noticed that the internal chat tool they'd built for their own distributed team was, actually, the thing people needed. They productized it. Named it Slack. Launched it in August 2013. Grew at 5 to 10 percent weekly.

Campfire had been there since 2006.

The question of why Campfire never became Slack is not really a mystery, and 37signals would tell you that plainly. They were not trying to build the next billion-dollar platform. They were trying to build useful software that made money without requiring them to become a different kind of company. Campfire succeeded at that. The fact that a different company later built a $27.7 billion business on the same idea is not evidence of 37signals' failure. It is evidence that the market for their idea was much larger than they were willing to pursue.

Whether that was wisdom or waste depends entirely on what you think software companies are for.

THE BREAKTHROUGH: "Rework" and the Anti-Manifesto

In 2010, 37signals published "Rework." It should not have been a bestseller. It was short, confrontational, and filled with advice that contradicted everything the startup industry believed. It became a New York Times bestseller.

The thesis of Rework — and of its predecessor "Getting Real," which 37signals originally sold as a $19 PDF before making it free, and which spawned a generation of indie hackers — was essentially this: growth is not the goal. The goal is building something useful, keeping it profitable, and protecting your ability to make decisions without answering to anyone.

They argued that workaholism was stupid — that working long hours was a badge of ignorance, not dedication. That meetings were toxic. That planning was guessing. That you should build less than you think, charge more than you're comfortable with, and hire slower than every piece of conventional wisdom tells you to. That the best thing you could do for your product was to say no — loudly, consistently, and without apology — to everything that didn't make it better.

Silicon Valley did not take this well. The ideology of that world — raise money, grow fast, dominate markets, worry about business models later — was precisely what 37signals was calling out. DHH in particular became a target and a hero simultaneously, the programmer who kept saying what programmers thought but weren't supposed to say: that "engineers" was a pretentious job title for people who wrote code, that VC money was a Faustian bargain, that shipping less was often better than shipping more.

They were not anti-business. They were anti a specific kind of business — the kind that required you to lose money for a decade in pursuit of market share, to hire hundreds of people you didn't need, to raise rounds that turned your company into someone else's asset. 37signals was proof that the alternative was real. Tens of millions in annual profit. A product used by millions. A team of fewer than sixty people. Zero outside investors.

For a certain kind of founder, they were a permission slip. You don't have to build a unicorn. You can build something smaller and keep it.

THE AFTERMATH: The Year Everything Fell Apart, and Then Kept Going

In April 2021, Jason Fried published an internal memo that he then made public.

The memo announced a series of policy changes at Basecamp — the company had renamed itself from 37signals to Basecamp in 2014 to signal its focus on a single product. No more societal or political discussions on company channels. No more DEI committees. No more lingering on past decisions. Benefits streamlined, the "paternalistic" ones eliminated.

The backstory was complicated. Since around 2009, some customer service employees had maintained an internal list called "Best Names Ever" — a collection of customer names they found funny, many of them Asian or African names, the kind of list that starts as a private laugh and calcifies into something uglier over time. A decade later, with the national conversation about race and corporate responsibility at a historic intensity, the list resurfaced internally and a DEI committee formed to process it. Fried and DHH, by their own later admission, had known the list existed for years and had not acted. When the committee's discussions became, in the founders' view, too politically charged for a work context, they shut the whole thing down.

The result was swift and severe. One-third of the company resigned. Not unhappy employees looking for an exit. Core people: the head of design, head of marketing, head of customer support — departments gutted in a single week. Basecamp offered severance to anyone who didn't want to stay under the new rules, and thirty people took it.

DHH later said the decision was one he and Fried stood behind. Fried, in a Bloomberg interview years later, cautioned others against banning political speech at work — not because he regretted the policy, but because the execution had been catastrophic and anyone considering the same move needed to understand what they were walking into.

The irony of the moment was almost too clean: a company that had spent twenty years publishing manifestos about how to run a workplace with integrity had a reckoning about how it had failed to do exactly that.

The company survived. HEY — an email client launched in 2020 with an aggressively opinionated design and a $99/year price point — had been a commercial success, signing tens of thousands of customers in its first months. In May 2022, citing their multi-product reality, they renamed the company back from Basecamp to 37signals. The circle closed.

Then, in January 2024, they launched Campfire again.

The new Campfire is not a SaaS subscription. It is part of a product line called ONCE — software you buy once, own forever, and run on your own server. $299, flat. One payment. Installs in one terminal command. Unlimited users. You get the code. You can modify it. There are no recurring fees, no per-seat pricing, no usage-based billing.

They called it ONCE #1. The direct answer to the question nobody had asked out loud: what if software worked the way software used to work, before the entire industry discovered that subscriptions were a better business than products?

They had built Slack before Slack. They had shut it down. They watched Slack get acquired for $27.7 billion. And now they were rebuilding it on their own terms, at $299, as a quiet rebuke to the entire model.

THE LEGACY: The Anti-Unicorn's Permanent Argument

37signals will probably never be worth $27.7 billion. They will almost certainly never be acquired. They will not IPO. They will not have a "growth at all costs" era, a "blitzscaling" chapter, or a round of layoffs after a valuation correction.

What they have built instead is something rarer in the tech industry: a company that made its choices consciously and lived with them fully.

Ruby on Rails changed how the web was built. Basecamp changed how small teams thought about project management. "Rework" gave a generation of founders a different vocabulary for success. Campfire proved you could build the future and still decline to own it.

DHH drives racing cars at Le Mans. Fried writes essays and runs one of the most profitable small software companies on earth. They employ fewer than sixty people. They answer to no board. They owe nothing to no one.

The Slack acquisition is worth $27.7 billion. The principles in "Rework" are still correct.

Both things are true.

5 THINGS NOBODY KNOWS ABOUT 37SIGNALS

1. Campfire launched seven years before Slack — and handled over a billion messages.
Campfire went live on February 15, 2006. Slack launched publicly in February 2014. In the intervening years, Campfire processed more than a billion messages for hundreds of thousands of users. It was not a failed product. It was a deliberately limited one. The company that later built a $27.7 billion business on the same idea was not doing something Campfire couldn't do. It was doing something Campfire wouldn't do.

2. Ruby on Rails was extracted from Basecamp as a side effect — and DHH released it for free.
DHH was not hired to build a framework. He was hired to build Basecamp. The framework he developed to build it was a byproduct. He open-sourced it in July 2004 and gave it away. The first Rails conference had 42 attendees. Within three years, Twitter, GitHub, Airbnb, and Hulu were all running on it. Over 1.2 million web applications have been built on Rails. None of that was the plan.

3. DHH won his class at Le Mans — five years after getting his racing license.
DHH took his first laps on a racing circuit in 2007, a hobby taken up with the same all-in methodology he applied to software. Five years later, he raced his first 24 Hours of Le Mans. In 2014, he won the GTE Am class driving a Gulf-liveried Aston Martin V8 Vantage as part of an all-Danish team with Nicki Thiim and Kristian Poulsen. He dedicated the win to Danish driver Allan Simonsen, who had died in an accident at the same race the year before.

4. The company renamed itself twice — first away from, then back to, its original name.
In February 2014, 37signals became Basecamp, shedding all its other products to focus on a single one. In May 2022, it renamed back to 37signals — because HEY had made them a multi-product company again, and they were, as Fried wrote, "headed back to a different time where we invented more and created more."

5. ONCE/Campfire is a direct philosophical attack on SaaS pricing — not just a product.
When 37signals relaunched Campfire in January 2024 at $299 one-time, they were not just shipping a chat app. They were making an argument: that the SaaS subscription model had extracted more value from customers than it had created, that the "pay forever for something you never own" mechanic was a structural harm to businesses, and that software could be sold the way software used to be sold. You pay. You get the code. You run it. You own it. ONCE is 37signals' thesis that the entire economic model of modern software is wrong — delivered as a $299 Ruby application that you install with one terminal command.

KEY FACTS REFERENCE

Research sources: 37signals/Signal vs. Noise blog, TechCrunch, NPR, Platformer (Casey Newton), 24h-lemans.com, Aston Martin Racing press releases, Ruby on Rails official history, DHH.dk, Changelog podcast, Wikipedia, Bloomberg.