Are AI Agents Really Driving the IT Freelance Boom? A Data-Driven Check

Testing the Rumour: AI Agents and the Rise of IT Freelancers

Have you been hearing this one lately?

"AI agents have got so good that IT freelancers are suddenly everywhere."

It has a science-fiction ring to it, which is part of the appeal — and it also invites the obvious question: is it really that simple?

In this article I want to work through:

• how much IT freelancing has actually grown
• how that growth relates to the spread of AI agents, meaning autonomous AI
• what the opportunities and risks look like if you want to make a living this way

using both the data and the story behind it.

---

Chapter 1: Are Freelancers, and IT Freelancers, Really Increasing?

1-1 Growth in Japan's overall freelance population

Start with the wider picture: how the freelance population as a whole has grown.

Domestic surveys put the freelance population at roughly 9.37 million in 2015 and 13.03 million in 2024 — an increase of about 39%.

In economic terms:

• roughly ¥14.64 trillion → roughly ¥20.32 trillion, also about a 40% increase

Against a backdrop of the side-hustle boom, work-style reform, and pandemic-era remote work, the number of people taking on work as individuals has grown not gradually but steadily.

The point to notice is this:

• the growth predates AI agents, and predates generative AI entirely
• and over a ten-year span, it is a large increase in absolute terms

1-2 Narrowing to IT freelancers

Now for the segment you are probably here for.

Estimates from agencies specialising in IT freelancers put the population at:

• around 100,000 in 2015
• around 160,000 estimated for 2025

That is roughly 1.6× over the decade.

Market size follows the same curve:

• roughly ¥719.9 billion → roughly ¥1,184.9 billion, again about 1.6×

More interesting still is the agency market that brokers IT freelance work. It is projected to expand about 3.9× over the decade from 2015 to 2025 — which tells you that the matching infrastructure between

people who want to go freelance

and

companies that want to use freelancers

has been built out substantially.

1-3 Globally, salaried employment is no longer the only option

This is not a Japan-only phenomenon.

• In the United States, roughly 40% of workers have done freelance work in some form
• Some forecasts have half of the US workforce freelancing by 2027

The global market for online freelance platforms more than doubled between 2018 and 2024, and is projected to keep expanding at roughly 17–18% a year through 2030.

Put roughly:

worldwide, more people are choosing to work as individuals rather than committing to salaried employment alone

Japan's growth in IT freelancing sits on top of that long-run trend, and it is natural to read it that way.

---

Chapter 2: Where AI Agents and Generative AI Actually Are

So how far along is the AI-agent side? Without pinning this down, you cannot tell whether freelancing grew because of AI or merely alongside it.

2-1 Enterprise adoption of AI agents

Global surveys report:

• 35% of companies have already deployed AI agents
• another 44% plan to deploy soon
• 76% of respondents feel AI agents are "closer to a colleague" than a mere tool

By AI agent here I mean:

AI that autonomously works through multiple steps, repeatedly judging and acting according to the situation

A step beyond a plain chatbot or one-shot question-and-answer generative AI.

A separate survey found 96% of CIOs saying they either already use autonomous AI or plan to within two years — so expectations on the corporate side are running high.

2-2 AI tool use among developers and IT freelancers

What about the developer side?

A global developer survey found:

• 51% of professional developers use AI tools daily
• 84% overall either already use them or plan to

Research on GitHub Copilot has reported:

• groups using Copilot completed tasks 55% faster than groups that did not
• around 90% of users said their productivity improved

Surveys narrowed to Japanese IT freelancers found:

• 63.3% use AI in their work routinely
• 24.3% use it occasionally

for a combined 87.6% using AI in some form.

The tools named include:

• ChatGPT
• GitHub Copilot
• Google Gemini
• Claude
• Microsoft Copilot
• Cursor, among others

and the uses run wide: technical research and information gathering, generating and suggesting source code, analysing error messages, debugging and refactoring, and writing documentation.

Another survey found 84.2% of IT and web freelancers using generative AI tools, with 54% of those on paid plans — evidence that a meaningful group is

actively investing in AI as a weapon

2-3 That said, AI agents still cannot do all of it

There is also data worth holding onto when you meet the "AI is taking the work" line.

One research team took real freelance jobs — listings from overseas crowdsourcing platforms — and set state-of-the-art AI agents on them. The reported result:

fewer than 3% of the jobs could be fully automated at a quality a freelancer could deliver

So as of now, the honest picture is closer to:

• an AI agent is about as capable as a quite good intern
• not yet a superhuman employee that replaces working freelancers wholesale

---

Chapter 3: How True Is "IT Freelancing Is Exploding Alongside AI Agents"?

With that material in hand, let us take the rumour's logic apart.

3-1 The timeline says AI is not the only cause

The chronology matters most.

• Japan's freelance population began growing in earnest around 2015–2016
• It grew further between 2018 and 2021, driven by pandemic remote work
• Generative AI such as ChatGPT, and AI agents, only became a live topic from 2022 onward

So neither

"freelancing grew, therefore AI appeared"

nor

"AI appeared, therefore freelancing grew"

is the right causal story. The natural reading is:

freelancing was already growing, and the AI-agent wave arrived on top of it

3-2 Why AI is nonetheless pushing IT freelancing along

Does that make AI agents irrelevant to IT freelancing? No — that would be going too far. There are three main ways they push.

① More AI projects in the first place

Reports on the IT freelance market show:

• listings for IT freelancers up 128% year on year
• listings in generative AI and data work specifically — prompt design, applying, evaluating, and improving large language models — more than doubling, up around 210%

Surveys on the corporate side add:

• about 59% of companies using IT freelancers for digital transformation deploy them in data and AI work
• about 70% of companies pursuing AI adoption cite a shortage of AI talent as a problem

The pattern is becoming common:

we want to bring AI in, but we do not have enough people internally, so we turn to IT freelancers

② AI tools arm you to fight alone

AI coding tools such as GitHub Copilot have been shown to:

• raise working speed by roughly 1.5× on average
• write routine code and tests automatically

Which produces a shift:

• system development that once needed a team of three or four is increasingly manageable with one or two people plus AI tools
• the ceiling on the scale and difficulty a solo freelancer can take on has risen somewhat

By analogy:

a sole trader who used to own only a light truck can now borrow a medium-sized one, for free, in the form of AI

Increase what you can carry — the size of the jobs you can take — and freelancing becomes a more realistic option.

③ Legal tailwinds and a stronger safety net

Japan's so-called Freelance Act took effect in November 2024.

In broad terms it aims to improve the conditions freelancers work under, through:

• an obligation to state the terms of a work contract in writing, email included
• a required payment deadline, in principle within 60 days
• seven prohibited practices, including unilateral fee reductions and refusal to accept delivered work
• obligations around harassment prevention and accommodating childcare and family care

Market data since then shows:

• applicants for IT freelance listings up 136% year on year after the Act took effect
• roughly 1.2× compared with before it took effect
• and still, over 90% of companies using IT freelancers say they intend to maintain or expand that use

In other words, efficiency gains from AI and a stronger legal safety net arrived together, and there is little doubt this has lowered the psychological barrier for people thinking:

maybe I will try IT freelancing rather than staying salaried

3-3 The verdict: right direction, wrong sole cause

Pulling the threads together:

• IT freelancer numbers and market size have grown roughly 1.5–1.6× over the decade
• work related to generative AI and AI agents has more than doubled in the last few years
• companies feel the AI talent shortage and are starting to place IT freelancers at the centre of digital transformation and AI adoption
• but the growth in freelancing began as a long-run trend before AI

So the claim

"IT freelancing is exploding alongside the rise of AI agents"

breaks down as:

• "IT freelancing is growing" → yes (the data shows substantial growth)
• "AI agents are the main cause" → no (changing attitudes to work, digital transformation, talent shortages, and legal reform all weigh heavily)

Half right, half overstated. Put another way, the realistic positioning is:

AI agents are supplying one more tailwind to an IT freelance market that was already growing

---

Chapter 4: The Upsides and Downsides of IT Freelancing in the Age of AI Agents

This may be the reason you are reading at all: what does choosing IT freelancing actually get you, and cost you?

4-1 Upsides for the freelancer

1. More high-rate work is available

In AI adoption and data work, areas such as prompt design, LLM evaluation, and AI architecture still have few qualified people, and average rates tend to run relatively high.

1. Even more freedom over place and time

With AI tools, research, drafting, and test-code generation can be turned around quickly from anywhere, at any time, which makes flexible arrangements practical:

• living in a regional area while working on Tokyo projects
• handling several clients' work efficiently

1. Easier to assemble a skill portfolio

Combinations such as:

• web application development × LLM work
• infrastructure × AI monitoring tools
• data analysis × automatically generated AI dashboards

let you build a core skill plus AI position, which makes differentiating yourself from other freelancers easier.

4-2 Downsides for the freelancer — or rather, realities

1. Headwinds for commodity work

In coding as in writing, work that consists of mechanically executing to a given specification overlaps with exactly what AI tools are good at.

That kind of work tends to see both:

• falling rates
• fewer commissions in the first place

which means the more you are simply selling hours, the harder this gets.

1. The load of continuous learning goes up, not down

AI tools and frameworks turn over every few months. Freelancers are more likely than employees to be expected to keep current on their own initiative, so the burden of self-investment — catching up between jobs, paying out of pocket to trial paid tools and learning material — arguably increases.

1. Income volatility and risk are still there

Surveys put average annual income for full-time freelancers somewhere in the ¥5–6 million range, but plenty of respondents also report months with zero income.

Even with AI on your side, you still have to keep thinking through risk management yourself:

• how to secure ongoing engagements
• what happens when illness or family circumstances cut your capacity

4-3 Upsides and downsides for companies

Worth summarising the corporate view briefly too.

Upsides

• fill gaps in AI and data talent on a spot basis, for exactly as long as needed
• try out people with sharp, specialised skills who are hard to hire as employees
• run experiments with generative AI and AI agents alongside outside specialists

Downsides and problems

• knowledge concentrates in individuals, and without handover it does not stay in the company
• the Freelance Act makes contracts, payment, and harassment prevention more complex to operate
• competition for AI-literate freelancers intensifies, making them harder to secure

So the corporate reality is less

AI cuts our labour costs, wonderful

and more a difficult puzzle:

how do we recombine the organisation around AI and freelancers

---

Chapter 5: How to Survive as an IT Freelancer in the Age of AI Agents

Finally, if you are either already freelancing in IT or thinking about going independent, here is where to put your attention.

5-1 Pick an axis of expertise: "AI × something"

From here on, the more valued profile is

an engineer who knows a domain deeply and can use AI well

rather than simply

an engineer who can use AI

Think in terms of taking a position at an intersection:

• B2B business systems × LLMs
• e-commerce × recommendation AI
• manufacturing × demand-forecasting AI
• internal knowledge search × AI agents

5-2 Treat AI agents as your subcontractors

From a freelancer's point of view, an AI agent is something like

a sub-member of the team who works around the clock

Work with that in mind:

• do the requirements yourself, then hand part of the implementation to an agent
• push test-case generation and code review to an agent, and keep your own attention on design and client conversations
• combine several agents — code generation, documentation, test automation — into a virtual team

Do this deliberately and you improve your odds of moving from having work taken by AI to using AI as staff.

5-3 Trust and repeat business are still human work

However far AI agents advance, some things stay on the human side for the foreseeable future:

• the interviewing that draws out what a client actually thinks
• the conversation about whether a given specification will really work as a business
• rethinking the framing of the problem itself

If anything, the more AI takes over coding and documentation, the more a freelancer's greatest asset becomes the trust-based relationship captured by:

give it to this person and the business side will actually work out

Which makes the decisive questions:

• how to design for repeat engagements rather than one-off jobs
• whether you can deliver value beyond the deliverable — proposals, improvements, teaching

That is where freelancers in the AI era win or lose.

---

Closing: The Conclusions, Headline by Headline

To recap the article at the level of its headings:

1. IT freelancing has grown roughly 1.5–1.6× over the decade — growth you can fairly call dramatic
2. But that growth sits on top of diversifying work styles, digital transformation, and talent shortages, all of which predate AI agents
3. AI agents and generative AI are pushing IT freelancing hard, in two ways: more AI-related work, and better armament for working alone
4. At the same time, freelancers doing commodity work face headwinds, and work is concentrating on those with expertise and trust
5. From here, the dividing line for survival is whether you can take a position at "business domain × AI" and use AI agents as staff

If this gives you the material to discuss the relationship between AI agents and IT freelancing structurally rather than sensationally, it should be genuinely useful to readers.