I Don't Use AI to Write My Code — I Use It as My Project Manager
Over the past year, AI has become a common tool among developers. Some developers use it to generate code, write tests, explain concepts, or even build entire applications from prompts.
My approach is a little different.
I don't primarily use AI to write my code. Instead, I use it as a project manager.
The Problem
One challenge many developers face is not writing code itself. The bigger challenge is knowing what to build next.
When starting a new project, there are usually dozens of decisions to make:
What should the architecture look like?
Which features should be built first?
What models are needed?
What can be postponed?
How should the project be organized?
What are the milestones?
As solo developers, we often play multiple roles at once: developer, designer, architect, tester, and project manager.
This is where AI has become extremely valuable for me.
How I Use AI
When I start a project, I explain the project requirements to AI in detail.
For example, I recently started building an e-commerce application using Django, HTMX, Alpine.js, Tailwind CSS, PostgreSQL, Redis, and Docker.
Instead of asking AI to generate the entire project, I ask it to act as a senior project manager and technical lead.
I provide:
Project goals
Features
Technical stack
Constraints
Business requirements
Then I ask AI to create a roadmap.
The roadmap is usually divided into phases such as:
Foundation and setup
Product domain
Cart and checkout
Payments
User accounts
Admin features
Deployment
After finishing a phase, I return and ask for the next phase.
AI then gives me a list of tasks that I can complete one by one.
Why This Works for Me
There are several reasons I prefer this approach.
- I Learn More If AI generates every file and function, I often find myself copying code without deeply understanding it. By implementing features myself, I stay involved in the technical decisions and understand how everything works.
- I Stay Organized Large projects can become overwhelming. Having a roadmap and a clear list of tasks removes the feeling of "What should I do next?" I simply focus on the current task.
- Better Project Structure AI can help identify missing pieces that I might overlook. For example: SEO requirements Inventory management Database relationships Deployment preparation Security considerations Having another "brain" reviewing the project plan can be very helpful.
- It Feels Like Working With a Team Lead As a solo developer, there is nobody assigning tasks or reviewing the overall direction. Using AI as a project manager creates a workflow that feels closer to working in a team. I still write the code, make the decisions, and solve the problems, but I have guidance when planning the next steps. The Downsides Of course, this approach is not perfect. AI can sometimes: Suggest unnecessary complexity Miss business requirements Create unrealistic roadmaps Recommend tools that don't fit the project Because of that, I never follow its suggestions blindly. I treat the roadmap as advice, not as instructions. Ultimately, I am responsible for the project. My Current Workflow My workflow usually looks like this: Define the project requirements. Ask AI to create a roadmap. Break the roadmap into phases. Complete one phase at a time. Return to AI for the next set of tasks. Implement everything manually. Review and adjust the plan when needed. This gives me the structure of a project manager while keeping the learning and coding experience in my own hands. A Question for Other Developers I'm curious how other developers feel about this approach. Most discussions about AI focus on code generation, but I rarely see people talking about using AI as a project manager, roadmap creator, or technical planner. Do you think this is a good way to use AI in software development? Would you trust AI to organize your projects and define your roadmap while still writing the code yourself? I'd love to hear how you're using AI in your development workflow.
Top comments (0)