Can you build reliable custom apps with AI? On Agentic AI Coding
Is it now realistic and worthwhile to build your own personal software? In software, the answer is increasingly yes, and tangibly so. Using Cursor, Claude Code, or a similar IDE, you can spin up a local or cloud-hosted application in minutes.
You can build tools with a flexible, modern architecture. The systems can be alive, fluid, adaptable, and centered around your specific workflows instead of rigid in their data structure. You build and you use, you learn and you iterate.
It’s akin to owning a house versus renting a house. Not taking into account financial implications, owning a house provides greater freedom than renting a house.
When you own a house, you can tinker with it, build small features, and work on small projects that make the house fit your vision increasingly more, providing a sense of joy and fulfillment. Similarly, owning software opens up space for such tinkering, playing, and experimenting to a greater degree than “renting” software from a SaaS (software as a service) company, where you are constrained by someone else’s functional and stylistic choices.
To make this concrete, I will build a local, personal tool to manage my work based on a workflow I have been refining and using for a few years. So far, I have built and implemented this workflow in Notion, which does a great job of providing building blocks for building tools. But what if I wish to make the app look and feel exactly as I desire, or go local and keep all the data on my local machine?
Workflows
Year Manifesto
The first viable scope for the app is a system to track and manage daily activities aligned with the broader life vision and objectives.
First, there is a Year Manifesto, detailing the vision for the year and for life. A year manifesto is a set of prompts I answer at the start of each year to reassess and crystallize where I am and where I want to go. Writing about your past, present, and future is a well-established method for increasing your chances of living the life you envision.
The process of writing itself is valuable for reflecting and clarifying what it is that you want. When you know what you want and lay out a vision for getting it, you significantly increase the chances of reaching the objective because you have now brought the thoughts and visions into life.
A manifesto is unlike a goal. A goal tends to be rigid and can become an obstacle to change. A manifesto is a work-in-progress draft of key principles for how you intend to live. It enables transformation. It doesn’t define milestones or set expectations. Rather, it sets the foundations for living in choiceless awareness, a state where you are in the driver’s seat, steering the boat that is your life.
Weekly Review
Every week, there is a process that consists of a review of the year manifesto, some questions to reflect on the week that passed, and some steps to lightly plan the week ahead. This weekly review is a tangible and frequent reminder of the vision for life, ensuring alignment remains between the direction of pursuit and the actions you undertake each week.
If you were to only do the year manifesto or goal-setting exercises, it is likely that a few weeks or months later, you'll have forgotten and fallen back to behaviours that don't align with the vision. So a weekly review is the frequent reminder that keeps vision and action aligned.
Projects and Tasks
These are work and life initiatives. Projects are “containers” of tasks. Tasks are single action items; small milestones to complete in the achievement of the project and, over time, of the year manifesto, if the projects are well-aligned with it.
A project-based approach to life and work provides space for experimentation and flexibility, as opposed to focusing excessively on following long-term goals. Once you know the general principles and direction, you can do the next smallest possible thing via a project. A project can be designing a website for a client, drafting a marketing assessment, finishing the 10 candles ordered by a customer, learning to juggle three balls, or learning basic plumbing.
There is a system of quick capture that allows you to easily add tasks from anywhere at any time, for later or immediate processing. This follows the quick capture concept popularized by Getting Things Done (GTD by David Allen).
Daily Activities
Each day, there is a constrained system for selecting projects and tasks (a maximum of 5 per day) to focus on that day. Throughout the day, that list stays front and center. You mark those activities as done along the way, so you gain useful insights over time. The next day, the list empties and a new selection occurs. This facilitates deep work and progress on all fronts of life, using a proactive rather than reactive approach for most of the day.
User Experience and Interface (UX and UI)
After using software tools for a long time, I notice some essential features that make a software tool worth using/investing in for me. Here are some of the main ones:
Fast and lightweight. It will load instantly and stay out of your way, facilitating focused work instead of hindering it.
Spacious and minimalistic, presenting one thing at a time, without any clutter on the screen. This reinforces the philosophy of focus and craftsmanship behind the workflows involved.
Flexible and scalable, also leveraging the latest tech (AI, vector DBs, local LLMs, embeddings) at its foundation for great flexibility and adaptability as the workflows evolve over time and the app gets shaped by them.
All the data stays private and easily exportable in bulk.
The app supports dark and light mode, allowing me to switch across modes easily.
The app works offline.
Implementation
As a first pass, I will build this app using Cursor, with a SQL + JSON approach for the backend (database). Cursor is a powerful ai-based programming tool, but I could use similar alternatives just as effectively (e.g., Claude Code, Antigravity). The SQL + JSON backend fosters flexibility in the data structure and it makes it easy to export data.
The app will only run locally when the computer is turned on, which I consider desirable because the nature of work remains context-dependent, like the craftsman’s workbench that lives in the workshop.
The SQL file will be stored in a Google Drive folder (indeed, this is an exception to going fully local), and the whole codebase will be synced to GitHub. This makes it easy to re-spin the app whenever I change computers.
Conclusion
Building your own custom software is possible now. The barriers to entry to this paradigm are increasingly low, and the whole software industry is shifting toward developer-grade tools that anyone can use. The question becomes: what are you building?
Want to build custom apps for you and your team? Submit an enquiry here.
Featured products
Resources
Pennebaker, J. W. (1997). Writing about emotional experiences as a therapeutic process. Psychological Science, 8(3), 162–166.
King, L. A. (2001). The health benefits of writing about life goals. Personality and Social Psychology Bulletin, 27(7), 798–807.
Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory of goal setting and task motivation: A 35-year odyssey. American Psychologist, 57(9), 705–717.
Gollwitzer, P. M., & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta-analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69–119.
Matthews, G. (2007). The impact of commitment, accountability, and written goals on goal achievement. Dominican University of California.
Duckworth, A. L., Kirby, T. A., Gollwitzer, A., & Oettingen, G. (2013). From fantasy to action: Mental Contrasting with Implementation Intentions (MCII) improves academic performance in children. Social Psychological and Personality Science.
Schippers, M. C., Scheepers, A. W. A., & Peterson, J. B. (2015). A scalable goal-setting intervention closes both the gender and ethnic minority achievement gap. Palgrave Communications.
