❓ Problem

Software is the only category where you can pay rental prices for twenty years and still own nothing at the end. Gartner has worldwide IT spending hitting $6.37 trillion in 2026, with software the second fastest growing line at over $1.4 trillion. SaaS specifically is forecast at $465 billion this year. Nobody is arguing that software is not worth buying. They are arguing about the invoice.

And the invoice is a mess. Zylo's 2026 index has the average organisation running 305 apps on $55.7 million a year of software spend, with only 54% of licences actually used. That is $19.8 million a year, per company, on seats nobody logs into. Down the market it is the same shape at a smaller scale: small and mid-size businesses typically run $50,000 to $500,000 a year on tools, and it hurts more because there is no procurement team to hide the bleeding.

Worse, the buyer has almost no leverage. Vendr's own framing is that skilled vendor negotiation causes most customers to overpay by 20-25%. Renewal season is a conversation between someone who does this twice a year and someone who does it four hundred times a quarter. It is not a negotiation, it is a sales call with a calendar invite.

Here is the kicker. The price of that software was set when building it meant three engineers and six months. That is no longer true. swyx watched a coding agent reconstruct a designer's entire website from a video in 43 minutes, unattended. Then his own team asked him to sign off on $40k a year for enterprise software they had never used, and the obvious question surfaced: could a weekend kill this?

Nobody can answer that question. The SMB owner has no time to run the experiment. The builder has no idea which SaaS is worth attacking. The two never meet, and the invoice renews.

✅ Solution

A bounty marketplace for killing your own SaaS, with a public index of the results.

• The business posts the target. The tool, the annual invoice, the ten workflows their team actually uses, and a cash prize. Not "build me a CRM", but "build the seven screens my ops lead touches every Tuesday, and here is the spec."
• Builders get a weekend. Standardised window, standardised scope, standardised eval. Everyone attacks the same brief, which is what makes the results comparable across the whole platform.
• The buyer's own team runs the eval. Not a panel of judges, not the platform. The people who use the tool daily score the clones against their real workflows, which is the only benchmark that matters and, conveniently, the only one nobody else has.
• Winner takes the cash, the code goes open source. The buyer gets a working replacement they own. The builder gets paid and gets a portfolio piece with a named enterprise attached.
• Then the winner gets first refusal on support. Building the clone is a weekend. Fixing it at 9am on a Tuesday when payroll depends on it is a job. That is the recurring revenue, on both sides of the marketplace.

The wedge is one angry invoice. The business underneath is the data exhaust: a live, public leaderboard of what got cloned in 48 hours versus what is still standing. Cloned in a weekend means the vendor was selling convenience. Survived twenty attempts means the vendor has a real moat, and probably deserves the renewal.

📊 Key Numbers

Market size

• TAM. The $465 billion global SaaS market, of which roughly 46% of licensed seats sit unused. That is not the addressable market, but it is the addressable anger, and it is worth well over $200 billion a year.
• SAM. Not everything is clonable. Strip out infrastructure, systems of record, anything with genuine network effects, and anything with regulatory or compliance lock-in. What remains is the single-workflow horizontal tier: form builders, scheduling, e-sign, approval flows, internal dashboards, reporting layers, lightweight project trackers. Call it 15-20% of SaaS spend, so a $70-90 billion pool of contracts a coding agent could credibly threaten.