How Cap makes money from open source software

This is the first post in a series where I look at how software companies make money.

I want to read pricing pages the way I read code. What is sold, what is free, what protects the paid part. Then I want to ask a more selfish question: what can we copy as indie developers, with no team and no funding?

I’m starting with Cap, an open source screen recorder. I picked it because it has the most visible tension possible. The full code is on GitHub. And there is a pricing page. Both things are true at the same time.

Prices and features below were checked on September 5, 2026.

What Cap is

Cap is a screen recorder for macOS and Windows. You record your screen and camera, and you get either a local video file or a share link.

It’s positioned as the open source alternative to Loom. Loom was acquired by Atlassian, costs $18 per user per month on its Business plan, and keeps every video on its servers. Cap says: same idea, half the price, and you own your data.

The desktop app is built with Rust and Tauri. The web app, where share links live, is Next.js. There is also a media server, a MySQL database, and S3-compatible storage. All of it is in one monorepo.

The project started in November 2023. Today it has about 21,000 GitHub stars.

The four things you can pay for

The pricing page has three plans plus a free tier. I’ll go through them from cheapest to most expensive, because each one sells a different thing.

Free

The free version is the full desktop app. You get the editor, 4K export, unlimited local editing.

Two limits. Recordings stop at 5 minutes. And it’s for personal use only.

That second limit is the interesting one. Nothing in the app checks whether you’re at work. It’s a rule in the license, not in the code.

Desktop License, $29 a year

This unlocks recordings longer than 5 minutes and gives you commercial usage rights. Their docs also list a $58 one-time option.

What does “commercial” mean here? Their licensing page defines it as recording anything for revenue-generating activities. Demoing your company’s product counts. Sending a video to a coworker counts.

So if you use Cap at your job, you owe them $29 a year. Nothing here touches their servers. You’re paying for permission.

Cap Pro, $12 per user per month

$8.16 if you pay yearly. This is the plan that looks like Loom.

You get unlimited cloud storage and bandwidth for share links. AI-generated titles, summaries, chapters and transcripts. A custom domain like cap.yourcompany.com. Password-protected shares. Viewer analytics. Team workspaces. A Loom importer to move your old videos over.

You also get the compliance badges: SOC 2 Type II, ISO 27001, HIPAA.

This is where the real business is. Every feature in this list costs Cap money per user. Storage, egress, transcription, inference. The subscription pays for that.

Enterprise, custom price

SSO and SCIM provisioning, SLAs, managed self-hosting, dedicated onboarding, volume discounts.

Same features as Pro, sold to a different buyer. The buyer here is a procurement department that needs a vendor to call.

There is also a developer API, billed by usage: about $0.05 per minute recorded and $0.001 per minute stored per month. It’s not on the main pricing page, but it’s a fifth revenue line.

Why giving away the code doesn’t kill the business

This is the part I wanted to understand.

The whole thing is public. You can clone the repo and run docker compose up -d, and you have your own Cap Web with MySQL, MinIO and the media server. Point the desktop app at it from Settings and you’re done. No 5-minute limit, no subscription.

So why does anyone pay? Three reasons, and each one is a deliberate choice.

The license is split in two

Look at the LICENSE file in the repo. It says two things.

The capture crates, the ones named cap-camera* and scap-*, are MIT. Anyone can take them and use them in anything, including closed source products.

Everything else is AGPLv3.

The MIT part spreads the project. Other Rust developers pick up the capture code, contribute fixes, and talk about Cap.

The AGPL part protects the product. AGPL says that if you modify Cap and run it as a service for other people, you have to publish your changes. That removes the scenario where a competitor takes the web app, rebrands it, and sells it as a hosted service. They could, but they’d have to open source their work too.

AGPL also makes a lot of companies nervous. Many corporate legal teams don’t allow AGPL code inside the building. That nervousness pushes them toward the paid plan, where the question doesn’t come up.

I wrote about the GPL family of licenses here if you want the background.

The commercial license covers the binary, not the code

This is a detail I almost missed.

The commercial license applies only to the app you download from cap.so. Their docs say it explicitly: it does not apply to versions you build yourself from source.

So the deal is: build it yourself and it’s free. Download the official build and use it at work, and it’s $29 a year.

Almost nobody builds a Rust and Tauri app from source to save $29. The license is unenforceable in theory and works fine in practice.

Self-hosting is offered, and it’s a lot of work

Self-hosting is real. But look at what you’re running: a Next.js app, a media server doing FFmpeg work, MySQL, MinIO, a reverse proxy with SSL, backups for two volumes. The AI features need your own API keys for transcription and a language model. And the docs say plainly they don’t offer support for self-hosted deployments.

For a screen recorder. Most teams will look at that list and pay $8 a month.

Self-hosting is not there to be used by most customers. It’s there so the customer knows they could. That’s what “you own your data” means in practice. The option changes how the hosted plan feels, even for people who never take it.

I did the same exercise on Plausible when I read the Community Edition code. Same pattern: real self-hosting, real hosted plan, and the hosted plan is what most people choose.

The 5-minute limit lives in open source code

One more thing I found interesting.

The free tier’s 5-minute limit is enforced in the desktop app. When you start a recording, the app runs a check. If the result is UpgradeRequired, it opens the upgrade window. You can read that code on GitHub.

In a closed source app, a limit like this feels like a wall. In Cap it’s a suggestion you can remove with a fork. Cap ships it anyway, because the people who fork it were never going to pay.

That’s a useful mindset. You don’t need to make the free tier impossible to bypass. You need to make paying easier than bypassing.

Who they are positioned against

Every sentence on the pricing page is aimed at Loom.

Loom Business is $18 per user per month. Cap Pro is $8.16 on the yearly plan. Loom keeps your videos. Cap lets you connect your own S3 bucket or Google Drive. Cap’s desktop app works offline. And Cap built a Loom importer so switching costs nothing.

Open source is the marketing channel for all of this. It’s the reason developers try it, tweet about it, and star the repo. Twenty thousand stars is a lot of free distribution.

Then the compliance work, SOC 2 and ISO 27001 and HIPAA, opens the door to companies that would never install a random open source tool. That’s the enterprise money, and it’s also the part a solo developer can’t copy.

What we know about the company

Cap Software, Inc. was founded by Richie McIlroy, based in Liverpool. He raised a pre-seed round to work on it full time. In mid-2025 he described the team as four people, fully remote. The repo uses Algora bounties to pay outside contributors for specific issues.

In June 2026 he posted an 82.8% month-over-month growth rate and said SOC 2 Type II was about to land. He doesn’t publish revenue, so I’d describe the size as early and growing, not big.

The pricing page says early-adopter prices are locked in for the lifetime of the subscription. That’s a signal prices will go up.

What I would copy

Now the selfish part. I don’t have a team, I don’t raise money, and I like it that way. I wrote about keeping a business small years ago and still believe it.

Here’s what from Cap’s model fits that constraint.

Free for personal use, paid for work. This is the cheapest paywall there is. No feature gating, no infrastructure. One sentence in a license file and a $29 checkout. The people who use your tool at a company have a card and an expense process. The people who use it at home don’t, and they were never going to pay anyway. Let them use it.

License the binary, not the code. If I released a desktop tool, I’d copy this exactly. Source on GitHub under AGPL. The official build behind a small yearly fee for commercial use. The build is the product. The source is the marketing.

MIT the parts you want spread, AGPL the product. Cap’s capture crates are useful to anyone writing a screen recorder. Those are MIT. The thing you’d actually sell is AGPL. This is a clean rule: the reusable library is free for everyone, the finished product has strings attached.

Pick one incumbent and price against it. Cap didn’t position as “a screen recorder”. It positioned as “Loom, but $8 instead of $18, and you own your data”. Every feature decision, from the importer to the custom S3 support, follows from that one sentence. If I were building a tool, I’d want to name the incumbent on my pricing page too.

Yearly, not lifetime, for the cheap tier. The Desktop License is $29 per year, not $29 forever. Small recurring revenue from a lot of people is more stable than one-time payments. I’ve written about recurring vs one-time revenue before, and Cap picks the recurring side even for a license that costs them nothing to deliver.

I’ve been on the other side of this too. When I made all my software free, it was because charging $20 per project added ceremony and nobody bought. Cap’s answer to the same problem is different: keep everything free to try, and charge only the people who have a business reason to pay. I think that’s the better answer when the product is something people run every day.

What I would not copy

The compliance stack. SOC 2 Type II, ISO 27001, HIPAA with signed BAAs. This takes months, a lot of money, and someone whose job is to maintain it. It only pays off if you sell to companies large enough to ask. Most indie products never will.

Unlimited cloud storage and bandwidth. “Unlimited” is a promise you can make when you have funding and a growth curve to justify it. Video is the most expensive thing to store and serve. As a solo developer I would price storage, or push people to their own S3 bucket, which Cap also supports.

Enterprise sales. “Talk to sales” means someone answers the phone, writes proposals, negotiates SLAs and fills out security questionnaires. That’s a job. If you’re one person, it’s your job, and you stop building.

Free self-hosting with no support. This works for Cap because the hosted plan is cheap and the self-host path is painful. If your product is easy to self-host, the same move eats your revenue. Plausible handles this by letting the Community Edition lag behind the hosted version. Cap handles it by making self-hosting a real infrastructure project. You need one of the two.

The pattern in one sentence

Give away the client and the code. Charge for hosting, AI, compliance, and the right to use it at work. Use AGPL so nobody can resell your hosted version.

It’s the same shape as Plausible, Cal.com and Ghost. Open source drives adoption and trust. The hosted plan pays the bills. The license protects the hosted plan.

What makes Cap worth studying is the $29 tier. It’s the one piece of the model that needs no servers, no team, and no funding. Just a license file and the understanding that most people would rather pay a small amount than build a Rust app from source.

That part, we can copy tomorrow.

Want me to talk about your product? You can sponsor this site.