What Does "Building with AI" Mean?
Lesson 7.1 — What Does "Building with AI" Actually Mean?
Most people who take an AI course are thinking about using AI, not building with it. But there is a whole spectrum between typing prompts into ChatGPT and being a software engineer building AI products from scratch — and much of that middle ground is genuinely accessible to people without any coding background.
This lesson maps out that spectrum clearly: what each level means, what you can realistically achieve at each level, who it is for, and how long it typically takes. By the end, you will be able to honestly assess where you are and where it makes sense to go.
The Spectrum: Four Levels
Level 1 — Using AI
What it looks like: Opening ChatGPT, Claude, or Gemini and having a conversation. Typing prompts and getting responses. Using AI features inside tools like Canva, Grammarly, or Notion.
Who it's for: Everyone. This is where all beginners start and where the majority of people — including highly sophisticated professionals — derive most of their value.
What you can achieve: Everything covered in earlier modules of this course. Writing, summarising, planning, learning, creative work, research.
Realistic timeline to proficiency: Days to weeks of practice.
What you don't need: Any technical knowledge.
The limit: You are operating within whatever interface the tool provides. You cannot connect AI to other systems, automate repetitive workflows, or create something others can use independently.
Level 2 — Customising AI
What it looks like: Creating custom GPTs or Assistants with specific instructions, knowledge bases, and behaviours. Setting up a Claude Project with a system prompt that keeps context across conversations. Building a chatbot on a platform like Poe.
Who it's for: People who want AI configured for a specific, repeated use case — their own work style, their company's tone of voice, a particular topic area.
Real examples:
- A lawyer creates a Custom GPT loaded with their firm's templates and common contract clauses
- A teacher builds an assistant with their course curriculum, so it only answers questions about their specific course content
- A small business owner creates a GPT with their product catalogue and FAQs, which they use instead of starting every customer query from scratch
Realistic timeline to first working customisation: An afternoon.
What you don't need: Coding skills.
The limit: The customisation lives inside an existing platform. You cannot easily share it as a standalone product with people outside that platform, and you have no control over what happens to the data.
Level 3 — Automating with AI
What it looks like: Using tools like Zapier or Make to connect AI to other apps and automate multi-step workflows. AI becomes a step in a process, not just a standalone conversation partner.
Who it's for: People who have repetitive tasks involving multiple tools — emails, spreadsheets, CRMs, databases — and who want those tasks to run automatically without manual intervention.
Real examples:
- Incoming customer emails are automatically classified, summarised, and routed to the right team member
- New form submissions trigger an AI to draft a personalised response, which a human reviews before sending
- A podcast's transcript is automatically processed to generate show notes, social media posts, and a newsletter summary
Realistic timeline to first working automation: A weekend for someone with no prior experience.
What you don't need: Deep coding skills. Zapier and Make are visual, drag-and-drop tools.
The limit: Complex logic, large data volumes, and custom integrations will eventually push you to code-based solutions.
Level 4 — Building AI-Powered Products
What it looks like: Using AI APIs directly to build applications — web apps, mobile apps, internal tools, and commercial products. This involves writing code (or working with a developer) and handling infrastructure.
Who it's for: Developers, technical founders, and increasingly non-technical people using AI coding assistants to build with minimal prior coding knowledge.
Real examples:
- A startup builds a customer service chatbot that integrates with their existing database, handles escalations, and learns from new conversations
- An individual creator builds a personalised content recommendation app for their community
- A business builds an internal tool that analyses sales calls, extracts key information, and updates their CRM automatically
Realistic timeline to first working product: Weeks to months. With AI coding assistants (GitHub Copilot, Cursor, Bolt.new), non-technical people are increasingly reaching this level — but it is genuinely challenging and requires significant investment.
What you need: Either coding skills, a technical co-founder/developer, or patience learning with AI coding tools.
A Map of the Terrain
| Level | Tools | Skill needed | Time to first result | Who it's for |
|---|---|---|---|---|
| Using | ChatGPT, Claude, Gemini | None | Minutes | Everyone |
| Customising | Custom GPTs, Poe, Projects | Basic prompt writing | Hours | Regular AI users |
| Automating | Zapier, Make, n8n | Logical thinking, no code | Days | Process-minded people |
| Building | APIs, Cursor, Bolt.new | Technical or patient | Weeks–months | Developers + ambitious beginners |
A Realistic Perspective on "No-Code AI Building"
There is a lot of enthusiastic content online suggesting that anyone can build sophisticated AI products with no technical knowledge in a weekend. Some of this is true. Much of it is optimistic.
Here is an honest view:
What is genuinely achievable by non-technical people:
- Simple chatbots with a specific knowledge base
- Automated workflows connecting 2–3 existing apps
- Personal tools for your own repeated tasks
- Basic web apps with significant AI coding assistant help
What still requires technical skill or a developer:
- Products that need to handle scale (thousands of users)
- Integrations with complex or legacy systems
- Applications requiring reliable data security and compliance
- Anything that needs significant custom logic
Key takeaway: The line between "no-code" and "requires code" has moved dramatically in the last two years. More is achievable than ever before. But "more achievable" does not mean "trivial." Set honest expectations and enjoy the process of learning rather than expecting instant results.
Which Level Is Right for You?
Ask yourself these questions:
-
Are you currently getting real value from Level 1? If not, go deeper there before moving up the stack.
-
Do you have a specific repeated task that would benefit from automation? If yes, Level 3 (Zapier/Make) is likely the right next step.
-
Do you want to create something others will use? If yes, you need Level 3 or 4 depending on complexity.
-
Are you comfortable learning through trial and error? Building with AI requires patience — things break, tutorials are sometimes out of date, and there will be frustrating afternoons.
The next lessons in this module walk through Levels 3 and 4 in practical detail. Let's start at Level 3.
Practice Task
Think about your own work or daily life. Identify one process that involves multiple tools and repetitive steps — something you do at least once a week. Write it down as a series of steps: "First I receive X, then I do Y, then I update Z, then I send A." That description is the seed of an automation. Hold onto it — you will use it in the coming lessons.