7 sposobów na ZARABIANIE NA AI w 2027

Executive summary

Michał Barczak presents seven ways to sell AI-enabled services, ordered roughly from easier to more technically demanding. His central business advice is to sell solutions to specific customer problems—not access to AI tools. Specializing in an industry and understanding its workflows can help differentiate a service, even in a competitive market.

The presenter says his company generates PLN 100,000–150,000 in monthly revenue, mainly from AI-based services. The estimates below are his assessments, not independently verified benchmarks.

Evaluation framework

For each opportunity, Barczak considers:

• Time to learn the basics
• Share of the work AI can perform
• Estimated tool costs
• Number of tools needed
• Competition, rated out of 10; higher scores indicate more competition

Seven ways to earn with AI

Opportunity Learning time Work AI may do Estimated tools and costs Competition 1. Clip long videos into short content About 7–9 days (the subtitles render this unclearly as “79 days”) Up to 90% Usually 1–2 tools; budget around $50/month 9.5/10 2. Copywriting and content repurposing 20–30 days 70–80% 2–3 tools; roughly $80–$100/month 8/10 3. AI-assisted websites and landing pages 30–60 days Around 80% 3–4 tools, plus hosting; budget around $100/month 7/10 4. Business-process automation About 90 days Up to 90% Around 5–6 tools; roughly $80/month as an estimate 6/10 5. AI agents About 3 months Around 70% to build and implement; agents may later operate autonomously 3–6 tools; around $180/month for a high-tier AI plan 6/10 6. AI avatars and video content About 15 days Up to 95% Usually 2 tools; budget around $500 for experimentation, with generation/API charges varying by use 2/10 7. Small, purpose-built applications At least 90 days Up to 70% 2–3 tools; likely requires a high-tier coding subscription, but no specific cost is given 3/10

1. Short-form video clipping

Turn creators’ long-form videos, podcasts, or interviews into short clips for social media. The speaker demonstrates Riverside, which can record and transcribe conversations, identify potentially interesting segments, and assist with editing and layouts.

He estimates AI can do most of the production work, but clients still benefit from someone who understands what content will serve their audience and business.

Business implication: Competition is high, so selling “editing” alone may be difficult. Position the service around a creator’s broader content and social-media needs.

2. Copywriting and content repurposing

Barczak argues that AI has reduced demand for low-value, generic writing, but has not eliminated the need for skilled copywriters. The opportunity is to specialize in a type of content, audience, or industry, and repurpose a client’s source material across channels.

He demonstrates Castmagic turning recorded meetings into possible articles, video titles, keywords, email copy, FAQs, and social posts. The tool can accelerate production, but understanding the client’s tone, audience, and marketing goals remains important.

Actionable recommendation: Learn basic marketing and copywriting, then specialize in a niche—for example, steel businesses, real estate, car dealerships, or beauty salons.

3. Websites and landing pages

AI can speed up website creation, but the speaker cautions that standalone website projects may be unattractive for a freelancer: acquiring the client can cost too much relative to the revenue from a one-off project. Larger agencies may make the economics work through bigger projects and dedicated sales and production teams.

He demonstrates a Hostinger integration that can generate and publish a simple landing page through an AI workflow. He gives PLN 1,500 per landing page as an example price and says several pages for a client could make the tool investment worthwhile.

The harder, higher-value work is understanding the client’s goals, structuring the page, choosing the right messaging and visuals, and using the page to test a business idea. He recommends specializing by industry rather than offering generic websites.

4. Business-process automation

Look for repetitive work that consumes employee time without creating much value. One example: 10 sales employees each spend 30 minutes a day copying data between spreadsheets or systems to prepare a report that a manager reviews only monthly.

The opportunity is to identify the waste, determine which systems can connect, and build a reliable workflow. Barczak emphasizes hands-on learning—building small automations—rather than only watching tutorials. He mentions low-cost self-hosting, including n8n on a VPS, as one possible approach.

Actionable recommendation: Start with one recurring workflow in a specific business or industry, quantify the time it consumes, and propose an automation tied to that cost.

5. AI agents

Agents can carry out defined tasks and interact with company data or tools. Barczak describes an agent built for an industrial and construction-sector client that could access financial data from multiple sources and let users query it through chat.

Potential applications he mentions include financial or e-commerce analysis, content distribution, market and pricing research, and checking email inboxes. The key challenge is not merely building an agent, but defining a useful, secure task and integrating it with the client’s systems.

Actionable recommendation: Build agents first for your own workflows, then adapt proven patterns to client needs. Start with a narrowly defined business task rather than a general-purpose agent.

6. AI avatars and video for businesses

Businesses that do not want to put an employee or owner on camera may still need a social-media presence. AI-generated avatars and videos can produce content at scale, particularly for online stores with many products.

Examples include stores selling handbags, jewelry, pet accessories, or children’s toys, where conventional video production can be expensive and businesses need frequent new product content. The speaker notes that costs are often usage-based—per image, video, or API tokens—rather than a simple fixed monthly subscription.

He recommends budgeting about $500 for a substantial learning and experimentation phase, then incorporating generation costs into client pricing.

7. Small, task-specific applications

The final opportunity is to create compact applications that solve a specific operational problem, such as calculators, quote-generation tools, or internal tracking systems.

Two examples:

• For a steel-industry client, a task that took technologists two weeks to price a single item was reduced to a few minutes with a purpose-built application.
• Another client received an application for checking employee permissions. It replaced scattered records in notebooks, calendars, and assignment tools, and let employees log in, view and update their information, with alerts handled through the system.

This work requires more technical and operational knowledge, including deployment, security, and code review. As with automation and agents, the differentiator is identifying a concrete business problem and designing an appropriate solution—not simply writing code with AI.

Cross-cutting playbook and recommendations

• Start with customer pain: Identify repetitive, costly, slow, or unreliable work before choosing a tool.
• Specialize: Focus on a specific industry or customer type so you can understand its language, workflows, and priorities.
• Sell outcomes, not tools: Explain the business improvement—time saved, content produced, or a faster process—rather than just the AI service used.
• Learn by doing: Build practical projects and test workflows instead of relying only on courses or tutorials.
• Treat AI as leverage, not a substitute for judgment: The presenter’s estimates for AI contribution range from 70% to 95%, but customer discovery, quality control, implementation, and business context still matter.
• Check project economics: Consider client-acquisition costs, recurring work, tool usage, and delivery effort. Barczak is particularly cautious about low-priced, one-off website projects.
• Price usage-based AI into the service: For avatar and video work, account for per-generation or API costs in quotes, or have the client pay those costs directly.

Metrics and figures mentioned

• Presenter’s company revenue: PLN 100,000–150,000 per month, mainly from AI services.
• Website example price: PLN 1,500 per landing page.
• Automation example: 10 employees × 30 minutes daily spent transferring data.
• Steel-industry example: quote preparation reduced from two weeks to a few minutes.
• Estimated competition is highest for clipping (9.5/10) and lowest for AI avatars (2/10), according to the speaker.

Presenters and sources

• Presenter: Michał Barczak.
• Tools and platforms discussed: Riverside, Castmagic, Hostinger, ChatGPT, Claude/“Cloud Code” (as rendered in the subtitles), and n8n.
• Companies and examples referenced: Hostinger (identified as a partner), Komputronnik (equipment collaboration), and an interview/podcast example involving Daniel from “Jan Nierdzewny.”