"AI × Cursor" — How to Get 10 Years of Skills in 10 Minutes: Writing Requirements Documents
"AI × Cursor" — How to Get 10 Years of Skills in 10 Minutes: Writing Requirements Documents
The era has arrived when you can build a full-scale application with zero programming experience. The skill of writing requirements documents — something engineers once spent ten years mastering — can now be completed in minutes through dialogue with AI.
In the first session of the engineer lecture series taught by Yoshiki Ando and Kazuki Naito, participants learn how to use Cursor — a tool that lets you code through dialogue with AI — to turn a business idea into a working application. For aspiring entrepreneurs facing technical barriers, AI is not just a tool — it becomes a powerful partner for turning ideas into reality.
This article covers the content from the first engineer lecture session.
Instructors
Name: Yoshiki Ando Affiliation: TIMEWELL, Technology Division GM / Principal Engineer
Name: Kazuki Naito Affiliation: TIMEWELL, Co-Founder and CTO
Contents
- Instructor Introduction
- Why requirements documents have changed in the AI era
- Basic Cursor setup and key notes
- Agent vs. Ask — how to use each and the @ reference feature
- Efficiently processing business documents in Markdown format
- A complete requirements document in just two prompts
- Choosing the right AI model and parallel processing techniques
- Purpose-built requirements definition services to explore
- Summary: A new way of building things with AI
Why Requirements Documents Have Changed in the AI Era
In the first engineer session, I'll share how to build full-scale applications using AI — even without programming experience.
The first thing I want to convey is the importance of requirements documents. A requirements document is documentation that clarifies "what are we building?" when developing a system — making it possible to give AI correct instructions. I spent ten years writing these at my previous job, and honestly it was quite difficult work. But not anymore. With AI, anyone can produce a first-rate requirements document without even realizing it.
A requirements document covers: project background, business requirements, user personas, functional requirements, non-functional requirements, and so on. Details like: "we need a login feature, and passwords must include both letters and numbers" or "screen transitions should complete within 1 second" or "the system should be designed to handle 10,000 simultaneous users" — all of this is documented.
Basic Cursor Setup and Key Notes
The tool we'll use in this session is called Cursor. Think of it as VS Code — the standard editor — rebuilt and integrated with AI to make it more useful. It's like a next-generation editor for coding that leverages AI heavily, essentially a coding-focused version of Notepad.
A few important notes about using Cursor. First, if you're on a company PC, there's a chance it may conflict with security policies — use a personal device if possible. Also, avoid using Cursor in folders synced to cloud services like iCloud or OneDrive — it can consume PC resources and cause unexpected behavior.
The most critical setting is Privacy Mode. Click the gear icon in the upper right, navigate to the Privacy Mode option at the bottom of General, and enable it. This prevents the code you write from being used to train Cursor's models. You must configure this within one day of installation or it automatically reverts — set it early.
Agent vs. Ask — How to Use Each, and the @ Reference Feature
For actual usage: Cursor has two primary functions, Agent and Ask. Ask is for simple questions — things like "what does 'platform' mean?" Agent is for when you want to edit code, and this is what you'll use the vast majority of the time.
Particularly useful is the file reference feature. Using the @ symbol, you can point to specific files or folders. For example, typing @input references the entire input folder. Dragging and dropping a file works the same way.
Looking for AI training and consulting?
Learn about WARP training programs and consulting services in our materials.
Efficiently Processing Business Documents in Markdown Format
In this session, we'll use the documents you created in the business lecture as input. Convert these to Markdown format and place them in the input folder. Markdown is plain text that supports formatting similar to HTML — it's a format AI can read easily. Open a Word file in Google Drive, then download it in Markdown format for an easy conversion.
A Complete Requirements Document in Just Two Prompts
Creating a requirements document actually takes just two prompts. The first prompt generates the overview; the second generates the detailed version. If the AI asks clarifying questions, respond interactively and refine — and you'll end up with a requirements document that says exactly what you intended.
For example, if you feel response time is slow, just say "response time feels slow, please make it as fast as reasonably achievable" — and the AI will suggest changing the initial screen load from 3 seconds to 1.5 seconds. That's how the back-and-forth refinement works.
Choosing the Right AI Model and Parallel Processing Techniques
Model selection matters. Click the gear icon, go to Models, and select Claude Sonnet 4. This model offers the best balance of cost and performance. There's a meaningful performance difference between versions 3.5 and 4 — if you can, subscribing to the paid plan produces better output.
Running multiple chats in parallel is also possible. Hold Shift and click the plus button to open a new tab and run multiple AI conversations simultaneously. I'm not great at multitasking so I don't use this much, but it's available when needed.
The requirements document we create today will be the input for the next session. Next time, we'll use a service called V0 to experience building an application quickly — you can say "make me an invader game" and a fully working game appears. That's the kind of service it is.
Purpose-Built Requirements Definition Services to Explore
There are services in the market — like GearIndigo — specifically designed for creating requirements documents. Paste your requirements document into them and they'll produce a clean, organized output that includes not just features but system diagrams as well. Just make sure to carefully review the terms of service before using any of them.
In the AI era, skills that engineers once took ten years to develop have become accessible to anyone. Use what's available, keep pace with the evolution, and build better and better things. Cursor isn't just for engineers. For example, we use it in mentoring sessions: putting attendance records and survey results into a CSV and loading it into Cursor to reference as we go.
Those who try it first win. Even if the questions feel trivial, keep asking the AI — getting comfortable is what matters. Session three and beyond is where we get into serious development, so get familiar with Cursor before then.
Summary: A New Way of Building Things With AI
This session showed how writing requirements documents — a skill engineers once spent ten years mastering — can now be completed in ten minutes through AI dialogue. Work that ten years ago had to be done painstakingly by hand now finishes in two prompts. This is not merely an efficiency improvement — it represents the democratization of making things.
Cursor is not simply an editor for writing code. It's a powerful partner for turning business ideas into reality. Using @ to reference files, knowing when to use Agent vs. Ask, and choosing the right model like Claude Sonnet 4 — with all of this, anyone can build a full-scale application without programming experience.
What matters is understanding that AI is not taking our work — it's extending our creativity. You don't need to understand the technical details in the requirements document. What's important is having a vision for what you want to build, and working with AI through dialogue to make it concrete.
Even non-engineers can turn their ideas into reality — if they have an idea. That's the new form of development in the AI era. Use what's available, keep pace, and keep building better things. That's the posture this era demands. From session three we get into serious development — first, become friends with Cursor. I'm excited to see your creativity expand infinitely with AI's help.
Inquiries and speaking requests →
Building Sustainable Competitive Advantage With "7 Powers" — The Seven Strategic Sources for Startup Success
Business ideas are everywhere — so why do so many startups fail? Even after rounds of customer interviews, even after building the product, they still lose to competitors. Many entrepreneurs face exactly this challenge.
The third session of the WARP program's business lecture covers the "7 Powers" framework for creating competitive advantage. This approach was developed by Hamilton Helmer, the strategist behind Airbnb, and systematizes the seven sources of sustainable competitive advantage for startups.
The 7 Powers framework provides a clear answer to this challenge. By examining your business through the lenses of scale economies, network effects, counter-positioning, switching costs, brand power, cornered resources, and process power, the path to sustainable competitive advantage becomes visible.
Lecturer
Name: Ryuta Hamamoto
Affiliation: TIMEWELL, Co-Founder and Representative Director CEO; Visiting Associate Professor, Shinshu University; Co-Executive Director, ONE X; CHANGE Lead, ONE JAPAN; Special CIO, Shiojiri City
Profile: Originally from Okayama Prefecture. Co-founded ONE X in April 2020 and serves as co-executive director. Has driven regional support for Shiojiri City, Ota Ward, and others; won the Work Story Award W in two consecutive years; received the Grand Prize at the Freelance Partnership Award 2023. Founded TIMEWELL in November 2022. Promotes the vision of building the world's #1 challenge infrastructure through tech and community. Has delivered 100+ lectures on generative AI and drives AI-driven development. TIMEWELL's SHIFT program is an approved Ministry of Economy, Trade and Industry course; WARP was selected for the Tokyo Metropolitan Government SUTEAM Program. Won the Excellence Award at the Yokohama Business Grand Prix 2024. Selected for Ministry of Economy, Trade and Industry J-StarX, SME Agency FASTAR, and Kanagawa Prefecture KSAP.
Contents
- Lecturer Introduction
- The seven sources of competitive advantage
- Switching costs and brand power
- Cornered resources and process power
- Using ChatGPT for strategic thinking
- Summary
The Seven Sources of Competitive Advantage
The first is Scale Economies. Think manufacturers and similar businesses — as production volume increases, average costs decrease, and competitive advantage grows with scale. For software and application services, this applies to infrastructure costs like AWS or GCP. There are costs you incur regardless of usage; the more users you have, the lower the cost per person. For businesses that combine hardware with software, every additional unit of hardware strengthens the scale economy dynamic.
One important caveat: falling into a labor-intensive business model means that no matter how hard you work, margins don't improve. Service-as-labor-substitution models are particularly prone to this, so understanding where you stand is essential.
The second is Network Economies (Network Effects). Think of phones or social media — the more users or participants there are, the more valuable the product or service itself becomes. A phone network with just two people is worth far less than one with three, four, or more — the value of connections grows exponentially.
Apple, for example, employs a model that captures both user network effects and developer network effects simultaneously. More recently, AI services where the system becomes more valuable as data accumulates are attracting significant attention.
The third, Counter-Positioning, means building a business model or strategy that existing competitors cannot easily imitate or follow. Think of Muji and Don Quijote — Muji cannot adopt Don Quijote's strategy. Their brand atmospheres are completely different. The Daiso/Seria relationship works the same way. Taking a position that competitors genuinely cannot easily copy creates its own barrier to entry.
Switching Costs and Brand Power
The fourth is Switching Costs — creating a state where customers, once locked into a product or service, find it very difficult to switch to something else.
I've personally poured significant money into the Apple ecosystem. Once you're accustomed to iOS or macOS, going back to Windows feels unthinkable. Once you've become skilled with a particular OS or application, the idea of learning something else feels deeply unappealing. Hardware tends to produce even larger switching costs. Fiber internet connections, once installed, become so inconvenient to change that people simply don't. Data, community ties, and accumulated proficiency are also important factors. For AI services, the more you use them, the more personalized they become — the system itself grows more aligned with you over time.
The fifth is Brand Power. For startups this can be difficult to achieve quickly, but brand trust, recognition, and the resonance people feel with your company's vision all contribute to competing on brand. The goal is to create a state where customers find so much value in the brand itself that switching becomes difficult.
This comes down to whether your brand story and marketing strategy are capturing the emotional and psychological investment of customers. "This service's story aligns with my values" is the state you're after. Muji handles its brand story exceptionally well — you find yourself buying things even though they're slightly more expensive. That's what a solid brand story produces. What is your brand story? Think about the story that only you can tell.
Cornered Resources and Process Power
The sixth is Cornered Resources — exclusively or preferentially holding human assets, intellectual property, or supply chains that are difficult for others to acquire. Patents are the obvious example, and signing an exclusive contract with the single authoritative figure in a niche means competitors can't access that person.
In a somewhat extreme case: Mercari secured investment from all the major VCs in Japan, effectively preventing competitors from accessing the same capital. By concentrating Japan's limited VC resources on Mercari, they created a situation where competitors couldn't raise funds — and won the market. There are also business deals where the game is about winning relationships with local governments — once you've connected with enough of Japan's approximately 1,700 municipalities first, it becomes very difficult for others to enter.
The seventh and final is Process Power — having proprietary know-how and operational processes that enable sustained cost reduction or quality improvement. The Toyota Production System and Tesla's automation expertise are classic examples.
What matters is whether your company has proprietary business processes, quality management methods, or operational optimization know-how that competitors find difficult to replicate. This is also an area that strengthens as you build the business. If you've already developed process power in a particular area from a previous role, bringing that into your startup or new venture is an option worth considering.
For example, deep expertise in security — the kind where you're confident no one can beat you at building secure products. Or exceptionally strong communication with traditional craft businesses — that's process power too. Customer communication know-how, stakeholder relationship-building, and partner communication capabilities can all count as process power.
Using ChatGPT for Strategic Thinking
When doing this kind of work, I recommend having ChatGPT open and working through it interactively. "I tried framing it this way — what do you think?" — using it in that mode deepens the analysis. ChatGPT might tell you something you hadn't noticed: "actually, data would definitely accumulate here."
Always do this kind of strategy work with ChatGPT, Claude, or Gemini open alongside you. In this era, there's no need to think alone. In my mind, it's as if I have an excellent advisor sitting with me — so use the advisor.
Finally, organize these ideas and, considering technical priority and ease of execution, find the areas within 7 Powers most likely to positively impact your business. If you can't come up with all seven — having one or two is more than enough in the early stage. If you have zero, they can be built. Working with ChatGPT, you can generate ideas: "this one, at least, might be achievable."
Use the 7 Powers framework to think about which to strengthen first, and start building your competitive advantage.
Summary
The 7 Powers framework is not mere theory — it maps a practical path to building competitive advantage. Scale economies, network effects, counter-positioning, switching costs, brand power, cornered resources, and process power — while each appears independent, they are actually interrelated and mutually reinforcing.
The key point: you don't need all seven from the start. Building a solid advantage in one or two areas first, then strengthening others as the business grows, is the realistic approach. Being careful not to fall into a labor-intensive business model, and clearly identifying where your strengths are — or where they can be built — is the first step toward success.
When thinking through competitive advantage, always use AI tools like ChatGPT. Conversing with AI uncovers new perspectives and possibilities you might have missed thinking alone.
7 Powers gives you answers to the fundamental questions: "why can we win?" and "how do we keep winning?" Use this framework as your compass, and build sustainable competitive advantage.
Inquiries and speaking requests →
TIMEWELL's AI Adoption Support
TIMEWELL is a professional team that supports business transformation in the age of AI agents.
Services
- ZEROCK: High-security AI agent running on domestic Japanese servers
- TIMEWELL Base: AI-native event management platform
- WARP: AI talent development program
In 2026, AI is moving from something you "use" to something you "work alongside." Let's think through your organization's AI strategy together.