Why Dan Koe Bets on One-Person Creative Teams as 2027's Top AI Business

As of 2026, AI allows a single person to research topics, draft articles, write reports, and repurpose long-form into short posts. However, if the saved time is simply spent taking on more client work, the work style may not become freer. Dan Koe's recent piece "The most lucrative AI business to start in 2027" identifies the "one-person creative team" as the prime opportunity. His core thesis: content attracts customers for businesses, and AI now lets one person shoulder research, strategy, and production that previously required a team. Someone who can organize these functions can serve clients who would otherwise need a full agency.

Two Paths

Koe outlines two routes: become your own creative team — building a brand, audience, email list, and products — or become someone else's creative team, producing content for founders, companies, or creators. Both require integrating brand positioning, content strategy, writing, and daily operations.

Why Now? Demand and Delivery

Content Carries Customer Acquisition

Koe starts from a simple question: no matter how good the software or book, if nobody knows who made it or what it offers, they won't pay. In his framework, content does the work of attracting potential customers and making the product understood. This capability is needed both for running one's own business and as a service for other operators.

A previous article ("Information Products Dead? The Solo Entrepreneur's Next Card") explored turning knowledge into interactive micro-tools. This new piece continues: once the work exists, how do you get people to know it and keep paying attention?

Demand doesn't mean founders have time to execute. Koe relays feedback from content agencies and media teams: clients are busy managing teams and companies, with little time to even think about content. Even with AI drafting, they still want someone responsible for strategy and creativity, applying tools to their specific business.

He also observes that many founders' content stays at the level of product announcements. News tells what happened, but doesn't make people continuously follow the person or understand their judgment. This leaves a service gap: helping clients extract what's worth saying from daily operations and experience, then turning it into sustained output.

One Person, Three Interlocking Responsibilities

The "one-person creative team" comprises three coordinated roles:

Brand Strategy — decides positioning and how others perceive the person. Koe warns that growth and engagement cannot come at the cost of trust and professional reputation. A viral hit can grab attention, but may not make readers want to keep following; consistent experience, stance, and trade-offs are what make readers remember who you are.

Content Strategy — researches what problems people care about, which formats, headlines, and narrative styles spread more easily. It helps decide what to write next, but must serve the brand direction. The key filter: "Even if this topic spreads easily, is it appropriate for this person to tell?" Chasing trends alone may bring traffic, but readers can't articulate why this person is worth following.

Writing and Operations — turns topics into posts, video scripts, long-form, or emails, and schedules publication. The writer operates under constraints from the first two roles: the expression must be something readers want to consume, yet align with the person's actual experience and stance. Handing a random idea to a model to generate a draft completes only a tiny fraction of the work.

AI is why Koe believes one person can now cover these responsibilities. He cites another category of agency feedback: AI raises individual output, letting teams take on more work without hiring. In his described workflow, AI helps research high-performing content, organize background, and draft; the human then reviews and rewrites. One person can therefore manage more stages, not just type faster.

Connecting both sides yields his rationale: clients lack time and content-organization ability, while tools expand what an individual can deliver. If one person can orchestrate strategy, production, and delivery, they can offer the service at a price below hiring a full team. The client buys someone to take content off their plate; the provider can potentially fulfill it with a leaner structure.

Background First, Content Second

Koe demonstrates the method with a "suppose you're my client" walkthrough. Solo brand operators can follow the same steps to organize background and plan content.

Interview to form an "intellectual signature" — the recurring experiences, stances, and judgment patterns that show up in a person's work. His questions are concrete: who do you want to help, what difficulties are they facing; what have you tried, what failed; who influenced you, and what common advice do you disagree with. These answers become a reusable brand-background document. Later, when selecting topics, knowing who this person cares for, what they can speak to, and why they judge that way gives AI context and gives the editor decision criteria. Its utility goes far beyond mimicking a few catchphrases.

Collect historical content — if the client has published before, gather 10–20 high-quality pieces per platform. Analyze core viewpoints, expressive traits, and recurring ideas into a second document. This accumulates the person's unique background to aid research and drafting, not to automate the entire writing process.

Research external content to build a "content map" — define a few main directions, then expand into specific topics, writable angles, and different perspectives. Others' high-performing content provides format and topic references; the first two documents determine how those references are re-expressed through the person's own experience and stance.

Thus research, topic selection, and drafting connect. AI can propose ideas and draft based on this background; the human reviews and rewrites. Koe aims for both a continuous topic pipeline and a recognizable personal voice, rather than hunting a fresh viral topic each time.

Why Clients Pay

Whether this becomes a paid engagement depends on what work the client is willing to hand over. The article examines a B2B founder and technical expert scenario, supplemented with research and a hypothetical dev-tool company.

Assume a dev-tool founder wants to help technical leads decide whether to adopt a tool. They could chase trends with "Top 10 Efficiency Tools," or draw from their own implementation experience to discuss why a successful pilot still led to abandonment, or why migration costs were underestimated.

The first topic, if disconnected from the founder's professional judgment, may get reads but won't help target buyers understand them. The second, backed by real records, lets readers gradually learn what this person values and under what conditions their advice is trustworthy. This illustrates Koe's distinction between brand and content, applied to a technical business.

One read isn't enough. Another article, a discussion, a specific trade-off — each shapes the impression. In the public intro to "How To Build A World," Koe links long-form, short content, and product, emphasizing long-term accumulation: people need something to keep consuming to gradually understand you.

However, being seen, being trusted, having a useful product, and closing a sale are different things. Trust can reduce some friction, but purchase still depends on urgency, fit, budget, and adoption cost. Attention is the start; the rest must be done piece by piece.

Why do companies pay to discuss professional problems? The Edelman & LinkedIn 2025 B2B Thought Leadership Impact Report (1,934 US respondents) offers clues. Among the "hidden decision-makers" subgroup — involved in procurement but not deeply versed in the product — 55% said reading vendor thought leadership (knowledge, guidance, unique views, excluding product-centric material) is part of their evaluation.

But buyers' willingness to read ≠ vendors' willingness to hire writers.

If the hypothetical company already has marketing channels and experts who can articulate the problem but simply lack time to write, an external service may help. Using the "pilot passed but team still abandoned" hypothesis, the interview would ask: where did adoption stall? What constraints existed, why was that decision made? Then obtain verifiable pilot records, migration checklists, or interview notes to judge what can be written and what still needs questioning.

"Improve R&D efficiency" is too broad; "pilot passed, why didn't the team continue using it" lets you dig into docs and supplement interviews. Existing content must also be cross-checked: if they claim to specialize in cutting-edge tech but their most concrete pieces are about team implementation, the experience, actual work, and planned topics must align for the content map to be useful. Non-existent experience cannot be hallucinated into the expert's name by the model.

Figure 1: From dev-tool adoption problem to verified long and short content

In the diagram, AI assists in organizing materials, comparing claims, and drafting; the service provider handles the probing, selection, and editing, then asks the expert to confirm facts, stance, and boundaries. Short content derived from verified long-form must retain original conditions — cannot delete "under what circumstances" just for spread.

The client may be paying for this complete delivery from raw material to finished product. If the client already has someone who can do these steps well and only lacks a drafting tool, the outsourcing rationale weakens.

AI affects both sides of the business. Providers can save time with it; clients can also use it to reduce outsourcing.

MIT News reported an experiment with 453 professionals: the group using ChatGPT completed assigned writing tasks ~40% faster. However, the tasks didn't require mastering a specific enterprise's or client's background, nor did they assess factual accuracy of output. That time saving doesn't directly translate to a cost reduction for the whole service.

A CESifo working paper (Demirci, Hannane, Zhu, 2024) analyzed a global freelance platform. In the eight months after ChatGPT's launch, job postings in easily automated categories (writing, coding) fell ~21% relative to more manual categories. This is early single-platform research, not a 2027 forecast, but it signals: doing it faster yourself may coincide with demand shrinking.

Therefore, fee viability still hinges on the concrete question: for which business types do you have deeper understanding, what can you ask and judge, and which work can you fully complete. Interviewing, researching, and editing will also be transformed by tools; no single step should be treated as permanently irreplaceable.

One Person's Delivery Scope

The public process of service provider storyarb serves as a reference: interview experts, produce long-form and case studies, then derive newsletters and social content.

However, storyarb assigns three roles — content strategy, writing, and customer success — and explicitly states they don't handle distribution. Even teams define delivery boundaries; a solo operator must be even clearer on what they can actually deliver. The existence of such a service supply doesn't prove a single person replicating it will be profitable.

Koe himself noted in a 2023 article that small teams already existed then. "One person" is a scale choice, not a claim of zero collaborators. The new piece also cautions that even with AI, running your own content, audience, and products can still be a full-time job.

Committing to multi-platform content, video, email, publishing, and growth simultaneously can fill all available time. If you save two hours drafting but spend two hours waiting for materials, confirming views, and handling last-minute revisions, net free time hasn't increased.

Starting out, you can focus on one expert, one audience, one primary channel: one interview, one long-form piece, then a few short pieces from the same material batch. Clarify upfront who provides materials, who reviews, to what extent, and who publishes — only then can you cost the work.

Koe suggests approaching ideal clients with a spec piece first. Without an interview opportunity yet, you can use public talks, articles, and product docs to demonstrate how you understand the business and select topics; unverified judgments are left for discussion.

The spec shows capability. A scoped paid trial then tests purchase intent.

Time for acquisition, research, interviewing, verification, communication, and revision must all be accounted for.

Whether the client renews and whether you can sustain the workload must be evaluated together. Revenue rising while hours grow faster may mean a business you don't want to continue.

Long-Term Accumulation in Service

Client work brings income and exposure to real problems, but that's not the whole of Koe's solo-business thinking.

In 2022 he warned that relying solely on outbound and referrals traps you in hard-to-scale client work; when you want to pivot, you may start from zero. Later, describing a starter path, he linked building your own brand, developing capability, offering small-scale services, and trying products — without requiring everyone to ship software first.

Viewed this way, the two paths in the new article relate more clearly. Doing content for clients helps you learn the business, but it won't automatically grow your own brand.

Writing "pilot passed, why didn't the team continue" for yourself means drawing from your own experience and research, confirming, and publishing under your name. Doing it for a client means extracting the client's experience, getting their confirmation, and producing work under the client's name. When the engagement ends, the provider gets revenue and delivery experience, but the client's followers don't become the provider's audience.

Figure 2: Same dev-tool topic, two routes' output and accumulation

The separate notebook in the diagram is what the provider must do themselves: turn effective interview questions into methods, continue researching problems with public materials, or showcase case studies with permission. These can improve the next delivery and seed the provider's own public work. The client's raw materials, identity, and content do not automatically become reusable assets for the provider.

For example, after several dev-tool deliveries, can you write your own analysis of tool-adoption challenges without disclosing client materials? Do recurring difficulties contain a piece that could become a clearer consulting offering, workshop, template, or tool? These can be tested gradually, no need to rush into a course or SaaS.

The earlier article "Information Products Dead? The Solo Entrepreneur's Next Card" also explored delivering experience in a different way. Koe envisioned embedding writing knowledge into small software that lets people learn, practice, and complete work simultaneously. He didn't say knowledge products would vanish, but tried to bring knowledge closer to actual use. This remains his judgment; whether it helps users must be tested.

From this angle, products and services can reinforce each other. Services reveal where users actually get stuck; public work lets others understand your judgment; tools can ship a portion of repetitive work. Every step needs feedback; accumulation doesn't automatically become revenue.

As founder of content tool Eden, Koe also promotes his product and course in the new piece. "Most profitable in 2027" is his business judgment; existing evidence doesn't prove a revenue ranking. But the "one-person creative team" is worth discussing beyond how many more gigs one person can take.

If you plan to try next year, start by noticing your next engagement: can your existing work help the counterpart understand your capability? When delivering a similar problem, do you understand it a bit more and spend a bit less on repetitive communication than last time? Those changes will determine whether this business can continue in the way you want.

References

Dan Koe: The most lucrative AI business to start in 2027 (https://x.com/thedankoe/status/2117664712532123680)

Leo Xiaolei: Translation of Dan Koe's 2027 piece (https://x.com/leo_xiaolei/status/2117779253660201239)

Dan Koe: The One-Person Business Model (How To Productize Yourself) (https://thedankoe.com/letters/the-one-person-business-model-how-to-monetize-yourself/)

Dan Koe: The One-Person Business Revisited (Turn Yourself Into A Business) (https://thedankoe.com/letters/the-one-person-business-revisited-turn-yourself-into-a-business/)

Dan Koe: How To Build A World (The 2-Hour Content Ecosystem 2.0) (https://letters.thedankoe.com/p/how-to-build-a-world-the-2-hour-content) — public intro only

Dan Koe: How I'd build a one-person business if I started over in 2026 (https://letters.thedankoe.com/p/how-id-build-a-one-person-business)

Edelman & LinkedIn: 2025 B2B Thought Leadership Impact Report (https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report)

storyarb: How it works (https://www.storyarb.com/how-it-works)

Zach Winn, MIT News: Study finds ChatGPT boosts worker productivity for some writing tasks, 2023-07-14 (https://news.mit.edu/2023/study-finds-chatgpt-boosts-worker-productivity-writing-0714)

Ozge Demirci, Jonas Hannane, Xinrong Zhu: Who Is AI Replacing? The Impact of Generative AI on Online Freelancing Platforms, CESifo Working Paper No. 11276, 2024-08 (https://www.ifo.de/DocDL/cesifo1_wp11276.pdf)

Code example

Dan Koe：The One-Person Business Model (How To Productize Yourself)（https://thedankoe.com/letters/the-one-person-business-model-how-to-monetize-yourself/）
Dan Koe：The One-Person Business Revisited (Turn Yourself Into A Business)（https://thedankoe.com/letters/the-one-person-business-revisited-turn-yourself-into-a-business/）
Dan Koe：How To Build A World (The 2-Hour Content Ecosystem 2.0)（https://letters.thedankoe.com/p/how-to-build-a-world-the-2-hour-content）。仅采用公开导言。