Running to Stand Still: The Freelancer Productivity Paradox

04 Nov 2026

04 Nov 2026 by Luke Puplett - Founder

I use AI every day in my work, and I'm not being paid a penny more for it. I don't expect to be.

That isn't a complaint. It's a question I can't put down: how is productivity in information work actually recognised? If the answer is money paid, then a freelancer is a small working model of the whole economy, and the model says something odd. I'll be expected to do more, what I build will let my client handle more for the same price, and because everyone else is doing the same, nobody ends up with much to show for it.

This follows on from my piece The Red Queen & The Infinite Work Machine. Here I want to walk through it slowly, with a toy example. The numbers are illustrative round numbers, not data.

Step 1: one freelancer

Say I charge £500 a day and work 200 days a year. That's £100,000. In that year I ship ten features for a client.

Next year, AI tools double my output. I ship twenty features in the same 200 days.

Can I put my rate up to £1,000 a day? I could try. But every other freelancer the client might hire has the same tools, so they can ship twenty features too, at £500 a day. If I double my rate, the client goes elsewhere. So I bill £100,000 again.

Officially, productivity is output per hour, and output is measured by its value in money. Same money, same hours. On paper my productivity growth is zero, while in reality I'm doing twice as much.

Step 2: one client

From my client's side, things look better at first. A feature used to cost them £10,000. Now it costs £5,000.

But their competitors hire freelancers with the same tools, at the same rates, so they get the same deal. Whatever advantage my client had is matched straight away. The saving doesn't sit in their profits for long. It goes into cheaper prices for their customers, or better products at the same price, because that's what it takes to keep up.

So their revenue barely moves either. They're shipping more, spending the same, and charging much the same.

Step 3: the whole economy

Now scale it up. If this happens across every firm buying information work, total spending on that work stays roughly flat. So do hours worked. GDP is built from spending, so measured GDP stays flat, and measured productivity with it.

The real gain is that each pound now buys twice as much. In principle, price indices are supposed to catch that by recording that the same money buys more. In practice, for software and services, they are poor at measuring quality. A better app at the same price looks like the same app at the same price.

So where does the gain show up? As more activity. More features, more versions, more meetings about the features, more work invented to fill the space that was freed. This is what I called the Infinite Work Machine: we don't bank the time, we spend it on doing more.

Step 4: who keeps it?

If the worker doesn't keep the gain, and the client competes it away, does anyone keep it?

Two groups, as far as I can see. The first is customers at the end of the chain, who get cheaper or better goods. That's a real benefit, even if nobody measures it well.

The second is whoever sells the scarce input. In this story that's the AI tools themselves, and the chips and computing power behind them. Every freelancer has to pay for them just to stay level with every other freelancer. Stop paying and you fall behind. That's the Red Queen: running as fast as you can to stay in the same place. The subscription is the cost of standing still, and it's revenue for the seller.

Almost none of the gain sticks to the person doing the work.

Objections

Won't rates rise eventually? Possibly, but only where something stays scarce. If the tools make everyone faster, speed stops being scarce. What can still command a premium is whatever the tools don't level out: judgement, trust, knowing a domain well, being the person a client already relies on. Rates may well rise for those things. They're unlikely to rise simply because output went up, because output is exactly the thing everyone now has more of.

Isn't this just deflation, which is good? Cheaper goods are good, and I'm not arguing otherwise. My point is narrower. If the benefit lands as lower prices and better quality, and price indices miss the quality part, then the official numbers will say productivity is stuck while life quietly gets better in ways nobody counts. That matters for how we judge whether AI is "working". It also matters to anyone selling hours, because lower prices for the customer are paid for by the seller's output getting cheaper.

Can't I just work fewer days? In the toy example, if I worked 100 days and shipped ten features, I'd earn £50,000. Selling time means that working less earns less, whatever the time produces.

Ways out: stop selling hours

The thread through all of this is that I sell time. When the price of time is set by the market and the tools are available to everyone, getting faster doesn't help me. It helps whoever buys my time.

So the way to keep some of the gain is to sell something other than hours:

• Fixed-price work. Agree a price for the outcome. If I deliver it in half the time, the difference is mine. The risk moves to me too, so I need to know my own costs well.

• Owning a product. Build something once and sell it many times. Being faster means building more, or building better, and the return isn't capped by a day rate.

• A share of the upside. Equity, revenue share, a bonus tied to results. If what I build makes a client more money, part of that comes back to me.

None of these is easy, and plenty of good work will still be sold by the day. But it's worth asking, for each piece of work, whether faster is my gain or my client's.

Where Zipwire fits

Zipwire is built around timesheets and invoicing, so I'm aware of the irony. But moving away from selling hours doesn't mean you stop knowing where your hours go. To price a fixed job sensibly you need a record of what similar work actually took, and you need the admin of getting paid to take as little of your own time as possible. That's what Zipwire tries to do for freelancers and agencies: keep a private time journal, turn it into timesheets and invoices with less fuss, and give you a clear record of the work you've done.

---

Learn more