>>
Technology>>
Artificial intelligence>>
How Technology Is Creating New...Every wave of technology produces new occupations. The printing press created publishers, editors, and typesetters. Electrification produced linemen, appliance repair specialists, and an entire commercial electrician trade. The personal computer gave us a software industry that did not exist as a recognizable employment category before the 1980s. None of this is new.
What is new is the velocity. Between 2020 and 2026, the global economy has produced more distinct job categories than the previous twenty years combined, and most of them share a common trait. They live inside platforms, depend on consumer-facing technology stacks, and resist classification by labor statistics agencies that were designed around factory work, retail, and professional services. The U.S. Bureau of Labor Statistics still publishes detailed data on millwright apprentices and roof bolt operators while struggling to assign a code to a person who earns six figures from short-form video monetization, live shopping commissions, and AI dataset annotation contracts running in parallel.
This gap matters for reasons that go beyond statistics. Traditional employment infrastructure assumes a roughly stable taxonomy of jobs. Universities build degree programs for known professions. Banks underwrite mortgages based on recognizable income categories. Tax authorities calibrate compliance around employer-reported wages. When the actual work people do drifts away from those categories faster than institutions can adapt, the result is not just confusion. It is an enormous opportunity for new business categories to emerge in the space between the worker and the platform.
This article maps the new categories of digital work, examines why the platforms that created them are structurally unable to support the workers inside them, and looks at the specialized agencies emerging as the operational layer for entire workforce segments. It closes with implications for investors, regulators, and the companies building infrastructure for what is becoming the fastest-growing slice of the global labor market.
Five categories of work now employ tens of millions of people globally and barely existed a decade ago. They share platform dependency, fragmented income sources, and a strong skew toward sole-operator income structures, but the day-to-day work in each is radically different.
The first and most discussed is the creator economy in its strict sense. People who earn the bulk of their income through audience monetization across YouTube, TikTok, Twitch, Instagram, Substack, Patreon, and equivalents. A 2025 SignalFire analysis put the number of full-time creators globally above 4 million, with another 200 million part-time. Goldman Sachs has projected the creator economy will exceed $480 billion in total economic activity by 2027. The category includes long-form video creators, short-form entertainers, newsletter operators, podcasters, and a long tail of niche-specific specialists, from finance educators to woodworking channel owners.
The second is platform-based interactive work, where the income comes not from passive audience metrics but from real-time interaction. Live shopping hosts on TikTok Shop, Whatnot, and Amazon Live. Video-based interactive professionals on platforms across Asia and Europe. Coaching marketplaces where the talent runs hundreds of live sessions per month. The defining feature is that the work is synchronous. Cancellations, dropped streams, or platform outages translate directly into lost income within a single session, which changes the operational risk profile dramatically.
The third is digital community management as a paid profession rather than a side function of marketing. The shift to Discord, Circle, Slack-based paid communities, and private Telegram groups created demand for full-time community operators, paid retainers for community-of-practice managers, and contract roles inside DAOs and creator businesses. A meaningful share of the people in this category started as unpaid moderators and migrated into formal income over the last three years.
The fourth, and the one most likely to surprise people who only watch the consumer side of the industry, is AI training and data curation work. Scale AI, Surge, Invisible, Outlier, and a growing list of competitors employ or contract over a million workers globally in roles that include reinforcement learning from human feedback, expert evaluation, red-teaming, and domain-specific dataset construction. Many of these workers earn rates that exceed traditional knowledge-work wages in their countries, and the category has scaled from effectively zero in 2019 to a multibillion-dollar segment by mid-2026.
The fifth is interactive video and live commerce more broadly. This includes live shopping in the consumer goods sense, live tutoring platforms, telehealth-adjacent informational services, and the rapidly growing category of video-based interactive professionals across multiple verticals. The work blends entertainment skills, sales skills, and platform fluency in combinations that did not exist as a single profession five years ago.
Why are these categories resistant to classification? Three reasons. They cross traditional industry boundaries, so a creator running a paid newsletter, a YouTube channel, and a coaching cohort is operating across publishing, broadcasting, and education simultaneously. They are platform-bound, meaning the underlying technology stack defines what the work looks like and changes its structure unpredictably. And they are sole-operator first, with most workers operating as solo businesses that handle production, marketing, finance, and customer support inside a single person's workload.
Labor statistics agencies are trying to catch up. The U.S. BLS opened a public comment period in 2024 on revising occupational classification to include creator and platform work, and the European Commission has done similar consultations. The institutional response is slow because the underlying job structures keep changing.
The platforms that enable these new job categories are technology companies, not workforce development companies. This sounds obvious, but it is the source of nearly every problem the workers in these categories experience.
A platform provides distribution, payment rails, and a feature set. It does not provide training in how to use the platform effectively. It does not provide career path mentorship from someone who has done the work before. It does not provide financial guidance to a person whose income arrives in unpredictable chunks from multiple revenue streams. It does not provide legal support when a contract dispute, copyright claim, or platform policy violation puts the worker's income at risk. It does not provide health insurance or retirement infrastructure. And it does not provide community in any meaningful operational sense.
The data on creator economy churn makes the consequence visible. Multiple independent studies, including ConvertKit's 2024 creator survey and a 2025 analysis from Linktree, put the early-stage churn rate for new creators above 90 percent within the first 12 months. The actual figure depends on how you define a creator, but the directional finding is robust across every dataset I have seen. Most people who attempt this work do not make it past the first year.
The reasons are not mysterious. New creators consistently report the same blockers. They do not know which platform to focus on for their specific content type. They underestimate the production time required to maintain a posting schedule. They do not know how to price their offerings or when to introduce paid tiers. They run into tax compliance issues they did not anticipate. They get demonetized or suspended for policy violations they did not understand. They burn out because they have no peer group to compare notes with. And critically, they have no way to evaluate whether their performance metrics are normal or whether they are signaling a real strategic problem.
Successful workers in these categories piece together their own support systems. They join paid mastermind groups, hire bookkeepers who specialize in creator finances, retain entertainment-adjacent attorneys, and form informal peer networks with other operators at their level. This is functional, but it is expensive, slow to assemble, and only accessible to workers who already have enough income to fund the infrastructure they need.
A useful frame here is that platforms unbundled traditional employers. Twenty years ago, a working professional received training, mentorship, benefits, legal coverage, financial planning support, and a peer group as part of their employment relationship. The platforms separated the distribution and payment functions from all of those support functions. The support functions did not disappear. The need for them did not disappear. They simply became the worker's problem to solve.
This is the gap that specialized agencies are filling.
Agencies are not new in any industry that has talent on one side and platforms or buyers on the other. Hollywood produced talent agencies in the 1920s. Modeling agencies emerged in the 1940s. Sports agencies arrived in the 1960s. The pattern is consistent. When the talent earns enough to make professional representation economically rational, agencies form.
What is new is the speed at which specialized agencies are emerging in digital work categories, and the operational scope of what those agencies do.
A creator economy agency in 2026 is not a booking shop. It is a workforce operator that may handle brand deal negotiation, contract review, tax structuring guidance, content strategy, platform-specific optimization, growth analytics, audience research, and increasingly, equity participation in the businesses its talent is building. Agencies serving live shopping hosts handle product sourcing, supplier negotiations, return logistics, and broadcast operations on top of the standard representation functions. Agencies serving AI training contractors run quality assurance, project scoping, and client matching at scale, often functioning more like staffing firms than traditional talent shops.
The revenue models split into three broad patterns. The first is the traditional percentage model, typically 10 to 20 percent of revenue generated, sometimes higher for hands-on managed services. The second is a flat retainer plus performance bonuses, more common in agencies serving high-volume operators or in segments where revenue is unpredictable on a per-month basis. The third is a profit-share or equity model, where the agency takes a smaller percentage but participates in the long-term enterprise value of the worker's business, a model that has migrated from talent management into creator economy operations over the last three years.
The economics work because the agency layer captures a meaningful piece of a worker's career-long value. A creator earning $200,000 per year who works with the same agency for five years generates somewhere between $100,000 and $300,000 in total agency revenue depending on the structure. Multiply that by a roster of fifty to two hundred operators and the unit economics become clear. The agencies that survive are the ones that demonstrably increase the worker's income enough to justify the take rate.
The category is no longer hypothetical. UTA and CAA have built creator economy divisions. Whalar raised growth-stage capital specifically to scale creator agency operations. Night, Underscore Talent, and a long list of dedicated digital agencies have built rosters in the hundreds. Live shopping has produced specialized agencies in the U.S., across Southeast Asia, and inside China that handle hundreds of broadcast hours per week on behalf of their talent. AI training is starting to generate its own agency layer, though the staffing-firm framing is more common there. Every emerging digital category eventually produces agency infrastructure, and the pattern is becoming predictable enough that founders are now entering specific subcategories with explicit agency-building intent.
For investors, the agency layer has an attractive property. It is largely uncorrelated with platform risk. A creator agency that loses a single platform can rotate its talent to other platforms. The platforms themselves do not have this option. They depend on creators staying on their specific stack. The agency, in effect, owns the relationship with the worker, and the platform is one of several distribution channels the agency uses.
Among the new digital work categories, video-based interactive professional work has produced agency infrastructure faster and more completely than almost any other vertical. The market size globally exceeds $9 billion in annual revenue based on conservative public-source estimates, with growth rates above 20 percent year over year through 2026. The work involves real-time interactive video sessions on platforms that take a percentage of revenue, with the talent receiving the remainder.
The reason agency infrastructure scaled faster here than in adjacent categories is structural. The work has a steep learning curve, which means new entrants without guidance spend the first 60 to 90 days underearning while they figure out which platforms suit their style, how to set up their workspace, how to price their time, and how to maintain a sustainable schedule. The platforms themselves provide essentially zero guidance on any of this. The income variance during the learning curve is high enough that a large fraction of new entrants exit before reaching sustainable income, which mirrors the broader creator economy churn pattern.
Agencies in this space discovered that compressing the learning curve was the highest-impact intervention they could make. By providing structured onboarding, platform-specific training, schedule optimization, and ongoing performance coaching, they reduced the time-to-sustainable-income from the typical 60 to 90 days down to a small fraction of that window. The unit economics work for both sides. The agency captures a percentage of revenue that, calculated against the income the talent would have generated solo, is more than offset by the higher income produced under agency guidance.
An example of this model in practice is CamStar's agency onboarding, which provides structured training and platform selection for newcomers. The agency approach reduces the typical 60-90 day learning curve to roughly one week of guided onboarding, with ongoing support through the first six months. The structural lesson generalizes. Agencies that demonstrably move new entrants past the initial learning curve faster than they would manage solo capture durable economics because the alternative for the worker is months of trial-and-error at significantly lower income.
The category is also instructive on the broader question of agency selection. Workers who research the agency layer before committing tend to outperform workers who join the first agency that approaches them. The reason is that the operational quality of agencies in any emerging category varies enormously, and the marginal economics of a good agency relationship versus a mediocre one are large enough to dominate most other career-stage decisions. This selection problem itself is creating a meta-layer of agency review and rating infrastructure, which is another business category that did not exist five years ago.
Across every new digital work segment I have studied for this article, the same dynamic shows up. Platforms create the category. Solo operators dominate the early period. The churn rate is high because the operational learning curve is steep. Agencies emerge to compress the learning curve and provide ongoing support. The agencies that scale are the ones that produce measurable income lift for their workers, and the category gradually shifts toward a hybrid model where agency-backed and solo-operator workers coexist with the agency-backed segment capturing a disproportionate share of total revenue.
This pattern is now repeatable enough that it should be expected in any new digital work category that emerges over the next decade.
The implications stretch across investment, workforce development, regulation, and how technology companies build platforms.
For investors, the creator economy infrastructure layer is one of the most defensible business categories that has emerged in the last decade. The agencies, the financial services firms building creator-specific banking and tax products, the legal services firms specializing in platform-bound work, the insurance products being built for variable-income knowledge workers, and the education companies producing certified training programs for new digital professions are all components of an infrastructure stack that scales with the underlying labor force. The growth math is straightforward. As long as the digital work categories continue to grow and as long as the support layer remains structurally undersupplied by the platforms themselves, the infrastructure businesses have a long runway.
For workforce development, the implications are uncomfortable for traditional institutions. Most degree programs do not prepare graduates for the work that the fastest-growing employment categories actually require. Some institutions are responding. The University of Southern California has a creator economy program. Several community colleges have launched short-form curricula in content production and audience development. But the gap between what institutions teach and what the new digital categories require is large enough that the operational training is increasingly happening inside agencies, inside online cohort programs run by individual creators, and inside platform-specific learning environments. This is a multi-billion-dollar workforce training opportunity that traditional education is mostly not participating in.
For regulators, the pressure to adapt tax codes, employment classifications, and benefits infrastructure to platform-based work is increasing. The U.S., U.K., and EU are all in various stages of updating worker classification rules. The most likely outcome is a hybrid category between employee and independent contractor that captures the operational reality of platform-bound work, with some portable benefits infrastructure attached. The agencies and infrastructure layer have a strong interest in this outcome, because portable benefits would reduce one of the largest sources of operational stress for their workers and would make the agency relationship more durable.
For technology companies building platforms, the lesson is that the support layer is not optional infrastructure. Platforms that ignore it produce high-churn workforces and lose ground to platforms that integrate at least some of the support functions natively. Several of the larger creator-focused platforms have moved in this direction already, building creator funds, partner manager programs, and structured education content. The platforms that have not are watching the agency layer capture margin that could have stayed inside the platform if it had been integrated earlier.
The broader strategic point is that the new categories of digital work are not a passing phenomenon driven by the pandemic. The pandemic accelerated the trend. The underlying drivers, including the maturation of consumer payment rails, the spread of high-speed mobile internet, the rise of platform-specific monetization tools, and the cultural acceptance of digital work as a legitimate career, are durable. The work is going to keep growing. The institutions that adapt to support it are going to capture a significant share of the value it produces.
A working professional in 2026 might earn their income from a combination of YouTube monetization, paid newsletter subscriptions, brand deal negotiations handled by a creator agency, AI training contracts run on a part-time basis, and equity in a content business they are building with two collaborators they have never met in person. None of those revenue streams existed in their current form fifteen years ago. The job category that describes this person does not appear in any government labor statistics dataset. And yet there are millions of people whose income looks roughly like this, with the number growing every quarter.
The companies building infrastructure for this workforce are themselves a new business category. They are not technology companies in the platform sense. They are not professional services firms in the traditional sense. They are workforce operators for entire labor segments that did not exist a decade ago. Some of them will become large independent businesses. Some will be acquired by the platforms whose gaps they are filling. Some will consolidate into multi-vertical talent operators that look like the holding companies of an earlier media era. The category itself is durable regardless of which corporate structures dominate.
The future of work is not a single trend. It is a continuously expanding set of categories that traditional employment frameworks cannot describe, supported by an emerging stack of agencies, financial services, education providers, and platform-adjacent infrastructure that is being built in real time. The companies that take the workforce-operator role seriously, with the same rigor that earlier generations of agencies applied to film, sports, and modeling, will define how this segment of the economy operates for the next two decades.