AI ile Para Kazananlar vs Atıp Tutanlar

Executive summary

The conversation argues that the most practical way to earn from AI is usually not to build a product first, but to use AI to solve a real customer problem as a service. Client work can build market knowledge, experience, and cash flow; a repeatable solution may later become a product.

The speakers emphasize combining AI tools with existing industry expertise, using agents to speed up familiar workflows, and keeping people responsible for reviewing quality, security, and customer experience. They caution against “get rich quick” claims and spreading effort across too many ideas.

Business playbooks and recommendations

Start with a problem, not a product

• Identify a specific problem you understand and can solve.
• Begin with a service or small project rather than investing heavily in a product with no proven demand.
• Look for recurring customer problems that might point to a product opportunity.
• Validate demand before committing to product development or scaling.

Apply AI to existing expertise

• Look for repetitive work in a field you already know, such as marketing, retail, logistics, or e-commerce.
• Use agents to handle routine tasks, then assess whether their output is correct and useful.
• Domain knowledge can help people notice opportunities that a technically advanced AI provider might miss.

Treat agents as supervised workers

• Give agents a defined task, review their output, provide feedback, and repeat.
• Expect iteration: agents may not complete complex work correctly on the first attempt.
• Use human judgment to spot errors, improve user experience, and decide whether a result meets the customer’s needs.
• Consider security and permissions; the guest says agent workflows still raise security questions.

Measure automation by business value

• Compare the time and cost of a task before and after automation.
• Prioritize workflows where AI reduces turnaround time or enables more work without sacrificing quality.
• Include model costs and the cost of errors or rework—not just subscription prices.

Focus rather than chase every opportunity

Avoid “shiny object syndrome”: having many ideas does not mean they can all be executed well at once. Choose one opportunity and give it enough attention to test and develop it.

Use communication as a business skill

Clear instructions, context, examples, and self-expression can improve AI outputs and support sales and marketing. The guest describes using a Socratic-style questioning method to examine assumptions and make strategic decisions.

Build resilience through action

The speakers frame rejection and setbacks as information and practice, rather than reasons to stop. One exercise they discuss is asking for a small discount in a low-stakes setting to become more comfortable negotiating. They also stress that entrepreneurship is not suitable or necessary for everyone.

Examples discussed

• Freelance work after a layoff: Doruk Yalçınsoy says he created an Upwork profile after being laid off. Within about a month, he was earning more than his previous full-time salary by providing business-development, marketing, and sales-automation services.
• Service-to-product progression: His company began as a service business, encountered recurring needs, and later developed a product. He says the team also trained and fine-tuned a speech-to-text model for a specific task.
• AI-assisted marketing operations: One example reduced a Meta advertising setup task from roughly an hour to a few minutes, with the agent doing the initial work and a person checking it.
• Digital product on Etsy: The guest describes using AI to create and list a Spanish-language digital wedding-planning spreadsheet. It reportedly made four or five organic sales without advertising. The product is delivered digitally, avoiding shipping operations.
• Internal workflow tool: He built a video-editing tool for his own use, collected user feedback, and improved it. The example illustrates starting from a personal problem and iterating based on feedback, rather than assuming a quick path to large-scale revenue.
• Industry-specific automation: A logistics example involved using an agent to answer recurring customer questions, such as container location and shipping prices.

Metrics and figures cited

These figures were reported in the conversation and have not been independently verified:

• Freelancing: Earnings reportedly exceeded the guest’s previous salary within about one month. He mentions an initial hourly rate around $15, but the later rate is unclear in the subtitles.
• Product expectations: The guest estimates that 99% of people expecting to build a product quickly and become millionaires will not achieve that outcome. This is an opinion, not a measured statistic.
• AI adoption: The guest estimates that fully integrated, reliable agent workflows represent less than 1% of current work.
• Time savings: A store-management task said to take 10–15 hours was described as taking about one hour with AI assistance.
• Task throughput: The guest says a Meta ad-group or campaign task that previously took about one to one-and-a-half hours could be completed in minutes with an agent, subject to review.
• Tool usage: A voice-to-text tool reportedly saved about 15,000 minutes over two months, according to the guest’s own tracking.
• AI subscriptions: The guest says cloud-based AI tools cost at least $100 per month in his setup; he also mentions using a $200 plan. These are personal tool costs, not general business benchmarks.
• API-cost risk: An uncapped map API combined with agent research reportedly generated a bill of about ₺15,000 for a single search. Set spending limits and monitor API usage.
• Etsy: The digital wedding-planner product reportedly made four or five sales organically, without ads.
• Pricing: The guest mentions offering a lifetime license for an app at $29.
• Unverified creator-income claim: The conversation mentions a US YouTuber said to earn $3 million per year using a clone. This is presented as an example, not substantiated evidence.

Key cautions

• Set API spending caps and monitor usage; agent-to-API workflows can create unexpected costs.
• Don’t equate a cheap or free model with lower total cost: errors and wasted time can outweigh subscription savings.
• Don’t automate tasks blindly. Review outputs, particularly where security, customer experience, or business decisions are involved.
• Avoid treating online income claims or flashy lifestyle marketing as proof of a repeatable business model.
• Keep meaningful human involvement in work where authenticity, judgment, or trust matter. The speakers expect live, personal content to retain value even as AI-generated content grows.

Presenters and sources

• Umut İLHAN — host of Umut İLHAN ile Sonraki Adım
• Doruk Yalçınsoy — guest and YouTube content creator; the main source of the business examples and AI-workflow claims discussed above.