From TIMEWELL
This is Hamamoto from TIMEWELL Inc.
With AI, You Can Now Build Real Applications With Zero Programming Experience
With AI, you can now build real applications with zero programming experience. Skills in requirements definition document writing that engineers used to take 10 years to acquire can now be completed in an instant through dialogue with AI.
In the first engineering course session — led by instructors Yoshiki Ando and Kazuki Naito — participants learn how to use "Cursor," a tool that enables coding through dialogue with AI, to transform business ideas into actually functioning applications. For anyone who is considering entrepreneurship but faces technical barriers, AI will be not just a tool but a powerful partner for turning ideas into reality.
This article introduces the content of Engineering Course Session 1.
Affiliation: TIMEWELL Inc., Technology Division GM / Principal Engineer
Profile: Originally from Tokyo. Passionate about robotics and AI research, achieved second place in the world at a home robot contest during university.
Worked in IoT development at a major electronics manufacturer, accumulating broad system development experience over 10 years. Led the specification design of a home management system for a residential district of more than 1,000 units and led the project to success.
Continued competing in hackathons and manufacturing events, winning the grand prize at the ONE JAPAN Hackathon x Tokai. Self-taught in cloud technology, independently launched and operated a site for social game players.
Looking for AI training and consulting?
Learn about WARP training programs and consulting services in our materials.
Hobbies Include Gaming and Otaku Culture Such as VTubers
Hobbies include gaming and otaku culture such as VTubers. Learns UI design from games and applies it to daily work.
Affiliation: TIMEWELL Inc., Co-founder and Director/CTO
Profile: Originally from Tokyo. Joined NTT Advanced Technology Corp. as a new graduate. At NTT Advanced Technology, engaged in SI business centered on cloud services. Subsequently served as Technical Lead and Scrum Master for new service creation and development at a business company, developing POS services, digital signage, and other services. As an independent activity, contributed to open-source software and developed and operated a job-change site for specific industries with peak 500,000 PV/month and average 300,000 PV/month. Currently focuses on design, development, and operation of the challenger assistant service "TIMEWELL," as well as new service development. Supports individuals, organizations, and companies who are excited but anxious about technology. Specializes in implementing applications that efficiently streamline operations at speed, emphasizing meaningful rewards and making full use of as-a-service approaches.
Instructor Introduction
- Why Requirements Definition Has Changed in the AI Era
- Basic Cursor Setup and Key Precautions
- The Agent vs. Ask Distinction and the Useful Reference Feature
- Efficiently Processing Business Documents in Markdown Format
- A Complete Requirements Definition Document in Just Two Prompts
- How to Choose the Optimal AI Model and Parallel Processing Techniques
- Requirements Definition-Specialized Services Worth Knowing
- Summary: A New Form of Development, Created With AI
Why Requirements Definition Has Changed in the AI Era
In this first engineering course session, we cover how to use AI to build real applications even without programming experience.
The first thing to convey is the importance of the requirements definition document. A requirements definition document is a document that clarifies "what are we building?" when developing a system, to give the AI correct instructions. In my previous job, I spent 10 years writing these — and honestly, it was no easy task. But that has changed. Using the power of AI, you can suddenly be writing first-rate requirements definition documents.
A requirements definition document covers project background, business requirements, user personas, functional requirements, non-functional requirements, and more. For example, noting that a login function is needed and passwords must include alphanumeric characters; that screen transitions should occur within 1 second; that the system should be designed to handle 10,000 simultaneous users — these kinds of detailed specifications are recorded.
The Tool Used in This Course Is Called "Cursor"
The tool used in this course is called "Cursor." This is a next-generation editor built on the existing VS Code editor, with AI integration for ease of use. Think of it as a coding version of Notepad — a next-generation editor that lets you code while making full use of AI power.
A few important notes when using Cursor: on company PCs, there is a chance it may conflict with security policies, so please use a personal PC if possible. Also, avoid using it in folders that are synced with cloud services like iCloud or OneDrive. This can cause excessive CPU/memory consumption or unexpected behavior.
An important setting is the Privacy Mode. Press the gear button in the upper right, select Privacy Mode at the bottom of General, and your own code will not be used to train Cursor. If you don't set this within one day of first installing it, it automatically reverts — set it early.
The Agent vs. Ask Distinction and the Useful Reference Feature
In actual use, Cursor has two main functions: Agent and Ask. Ask is used for simple questions. For example, you can ask "What does 'platform' mean?" Agent, on the other hand, is the feature you use when you want to edit code — and in practice this is the one you will use most.
Particularly convenient is the file reference feature. Use the @ symbol to specify the file or folder you want to reference. For example, typing "@input" references the entire input folder. You can also reference files by drag and drop.
Processing Business Documents in Markdown Format
In this course, participants use the documents they created in the business course as input. Convert these to Markdown format and place them in the input folder. Markdown is a plain text format with HTML-like formatting that AI can read easily. You can easily convert by opening a Word file in Google Drive and selecting Markdown format for download.
Creating a requirements definition document is actually completed with just two prompts. The first prompt creates an overview; the second creates a detailed version. If the AI asks questions, refine it through dialogue, and a requirements definition document that matches your vision is complete.
For example, if response time feels slow, instructing "The response time feels slow — please make it as fast as generally possible" will have the AI propose changing the initial screen display from 3 seconds to 1.5 seconds. Refine through dialogue in this way.
How to Choose the Optimal AI Model and Parallel Processing Techniques
Model Selection Is Also Important
Model selection is also important. Press the gear button, select Models, and choose Claude 3.5 Sonnet — recommended as the best balance of cost and performance. There is a performance difference between 3.5 and 4, so paying for the subscription where possible will yield better output.
Running multiple chats in parallel is also possible. Hold Shift and click the plus button to open tabs and proceed with multiple AI dialogues simultaneously. I am not very good at multitasking so I rarely use this, but it can be leveraged when needed.
The requirements definition document created today becomes the input for the next session. Next time, participants will use a service called V0 to experience building an application easily. It is a service where just saying "build me an Invader game" produces an actual working game.
Services Specialized in Requirements Definition
There are also services in the world like GearIndigo that specialize in creating requirements definitions. Pasting the requirements definition document you have created will neatly compile not just functions but system diagrams as well. However, make sure to carefully review the terms of service before use.
In the AI era, skills that engineers used to take 10 years to acquire have become available to everyone. Use what is available and build progressively better things at speed. Cursor can be used by non-engineers. For example, we use it in mentoring sessions — putting participants' attendance status and survey results into CSV and feeding it to Cursor to reference as we proceed.
There is an advantage to those who try it first. Ask AI questions — even "dumb" ones — and get comfortable with it. From Session 3 onwards we move into serious development, so please get familiar with it by then.
This lecture has conveyed that writing requirements definition documents — a skill that engineers used to take 10 years to acquire — can now be completed in just 10 minutes through dialogue with AI. The work that we struggled with by hand 10 years ago can now be finished with just two prompts. This is not merely an efficiency improvement — it means the democratization of making things.
Cursor Is Not Simply an Editor for Writing Code
Cursor is not simply an editor for writing code. It is a powerful partner for turning business ideas into reality. By referencing files with @, switching between Agent and Ask appropriately, and selecting the right model such as Claude 3.5 Sonnet, you can build serious applications even without programming experience.
The important point is that AI is not something that takes our jobs — it is a tool that expands our creativity. You don't need to understand the fine technical requirements of a requirements definition document. What matters is having a vision of what you want to build and working through AI dialogue to make it concrete.
Even non-engineers can turn their ideas into reality — as long as they have the idea. That is the new form of development in the AI era. Use what is available and keep building better things at speed. That is the attitude required in today's world. From Session 3 onwards we move into serious development — but first, please make friends with Cursor. I look forward to seeing your creativity expand infinitely through AI.
Related Articles
- The Reality of a Working Mother Returning from Two Maternity Leaves — And How Her Work Philosophy Changed | TIMEWELL
- Before Taking Paternity Leave (Part 2): Three Absolute Must-Dos for Taking Leave During a Busy Season
- A First-Class Architect Who Stays Close to the Job Site — Finding My Own Path as the Fifth-Generation Leader of a Construction Company