Choose your business model
7 models that work with agents — and how to pick one.
Goal: choose a concrete, achievable business model in which an AI agent genuinely adds value —
and validate it before you start building.
The golden rule
Choose the business first, then the agent. Not the other way around.
People who start with "I want to do something with AI agents" often build something technically impressive that nobody wants to buy. People who start with "who has a painful, repetitive problem I can solve?" make money. The agent is a tool, not a product.
7 business models that work with agents
Each model includes: what it is, who it's for, the agent's role, and how you earn.
1. Productized service (service as a product)
You sell a clearly defined service at a fixed price, and the agent does the lion's share of the work.
- Example: "SEO product descriptions for online stores — $2 per product."
- Customer: e-commerce stores, marketers.
- Agent's role: generates, optimizes, and delivers the copy.
- Revenue model: per unit or monthly subscription.
- Why it's great: well-defined, repeatable, easy to price. Best starting point for beginners.
2. Content engine
You build a publication (newsletter, niche site, social account) that the agent fills, and you earn from ads, affiliate commissions, or subscriptions.
- Customer: readers/followers + advertisers.
- Agent's role: researches, writes, schedules, rewrites.
- Revenue model: affiliate, ads, paid subscribers, sponsorships.
- Watch out: distribution is everything; pure AI content without a genuine angle stands out (and
is sometimes penalized by search engines). Add human editorial judgment and a real niche.
3. AI-assisted freelance / agency
You offer a classic service (copywriting, research, data work, support) but deliver faster and cheaper because the agent handles the heavy lifting.
- Customer: small and medium businesses, other agencies.
- Agent's role: does 80%, you handle quality control and client contact.
- Revenue model: hourly rate or project fee (and your margin grows as delivery gets faster).
- Why it's great: you can have clients today and the risk is low.
4. Micro-SaaS / API service
You build a small software product where the agent is the core, and you sell access.
- Example: "Paste your meeting transcript, get action items + follow-up emails."
- Customer: professionals, teams.
- Agent's role: is the product (the processing behind the scenes).
- Revenue model: monthly subscription, pay-per-use.
- Watch out: more technical work required (users, payments, hosting). Module 09 helps.
5. Marketplace operator (e-commerce / dropshipping support)
An agent manages listings, prices, inventory, customer questions, and orders for your (or someone else's) online store.
- Agent's role: inventory management, listing optimization, support handling.
- Revenue model: your own sales margin, or a fee for managing someone else's store.
6. Lead generation & follow-up
An agent finds, qualifies, and follows up on leads for sales organizations.
- Customer: B2B sales teams, real estate agents, consultants.
- Agent's role: researches prospects, personalizes messages, follows up, schedules appointments.
- Revenue model: per qualified lead, per appointment, or retainer.
- Watch out: spam and privacy regulations. Do this properly and with consent — module 11.
7. Selling internal automation (consultancy)
You build agents that automate the internal processes of a business, and charge for the build and ongoing maintenance.
- Customer: small and medium businesses with repetitive administration.
- Agent's role: custom-built per client.
- Revenue model: build fee + monthly maintenance. Highest margins, but requires sales skills.
Which model fits where you are right now?
──────────────────────────────────────────────────────────────────
LOW TECH BARRIER HIGH TECH BARRIER
│ │
▼ ▼
[1] Productized [3] AI-assisted [4] Micro-SaaS [7] Consultancy
Service Freelance / / API service
(best for day 1) Agency
LOW SALES EFFORT HIGH SALES EFFORT
│ │
▼ ▼
[2] Content [5] Marketplace [6] Lead gen [7] Consultancy
Engine Operator
💡 In Claude.ai: Describe your business idea in a Claude.ai chat and ask it to stress-test your model against the five evaluation criteria below. It will spot gaps you might miss — and it's free to try before committing to any build.
How do you choose? The evaluation matrix
Score your top-3 ideas on these five questions (1–5 per question):
| Question | What you're looking for |
|---|---|
| Pain | How much does the customer currently suffer from this problem? (high = good) |
| Frequency | Does the problem recur often? (frequent = good for automation) |
| Willingness to pay | Do people already pay for a solution? (yes = proven market) |
| Agent fit | Can an agent genuinely do this well? (see module 00, the "can do" list) |
| Reach | Can you actually reach these customers? (your network/channel) |
The idea with the highest total score is your starting point. Pay extra attention to willingness to pay and reach — those are where most beginners stumble.
Validate before you build (critical)
Don't build first and then look for customers. Flip it around:
- Talk to 5 potential customers. Describe your service. Ask: "Would you pay for this, and
how much?" Listen to their objections.
- Do it manually first (or semi-manually with individual prompts). Deliver the service once
by hand. This teaches you the process you'll automate later — and you already have a client.
- Ask for a pre-commitment or deposit. Only when someone actually hands over money or a firm
commitment do you know it works.
Only once this succeeds do you build the agent. This saves you weeks of building something nobody wants.
Your assignment
- Write down your top-3 ideas.
- Score them on the evaluation matrix.
- Choose the winner and phrase it in one sentence:
> "I help [who] with [what problem] by [what the agent does], and I charge > [pricing model]."
- Fill in the business plan template.
Keep that one sentence. It's the compass for the rest of the course.