Building PositionTracker With AI-Assisted Development

After years of running publishing systems, I started building PositionTracker: a dashboard for organizing and analyzing stock and options positions. The product came from a practical problem — positions, research, and market context lived in too many places.

Individual investors can see quotes, options chains, filings, and notes in separate tools. Spreadsheets do not keep up. I needed a product that made the book of positions understandable without pretending to be a brokerage or a source of financial advice.

Founder. I directed product concept, user workflows, interface design, and implementation using AI-assisted development. I review, test, and decide what ships. I do not claim to have hand-coded every line or to be a senior software engineer.

• Define the job: one place to see positions, context, and the next question a user should ask.
• Choose a stack that supports authentication, financial data, a real database, billing, and frequent iteration.
• Use AI coding agents to move faster on implementation while keeping humans responsible for product judgment and quality.
• Apply publishing lessons: information architecture, clarity, and measurable outcomes.

• Designed workflows and a responsive interface around portfolio and options use cases.
• Integrated financial data, Clerk authentication, Neon Postgres, Stripe subscriptions, and Vercel deployment.
• Worked through GitHub, pull requests, testing, and release habits rather than treating the app as a one-off prototype.
• Kept the public claim accurate: this is a live product I built and operate, not a claim of institutional asset management.

• AI-assisted development with human review
• Auth, data, billing, and hosting as first-class product pieces
• Subscription plans and user-account workflows
• Iterative release on Vercel

• A live SaaS product at positiontracker.trading.
• A working example of moving from content operations into product operations.
• A clearer view of how content, software, and business models inform one another.

• AI agents change the pace of software work. They do not remove the need for taste, testing, or accountability.
• Product work makes publishing instincts more concrete: users either understand the interface or they leave.
• The transferable skill is systems: define the job, choose the constraints, measure whether the thing works.