A private Gallery Manager connects to Adobe Lightroom, discovers albums and folder structures, synchronizes additions, removals, ordering, captions, dates, locations, and cover images, and publishes organized galleries to the public site.
Impact / Applied AI
From portfolio
to product.
I used an AI-native, agentic product-development workflow to turn my personal website into a production digital platform—combining portfolio storytelling, photography publishing, social content, contextual AI assistance, and API-connected management tools.
2Custom publishing managers
3 APIsLightroom, Threads + OpenAI
Human-ledReview before publishing
LiveProduction platform
The opportunity
Demonstrate applied AI through a working product
I did not want my website to simply state that I understand AI strategy. I wanted it to provide evidence: a useful, evolving product that shows how I identify opportunities, design workflows, connect systems, set guardrails, and move from an idea to working software.
That changed the brief from “redesign a portfolio” to “build a small digital platform”—one that I could continue operating and improving myself.
The working model
Direct agents like a multidisciplinary product team
I set the product vision, define the user problem, establish requirements, make experience and content decisions, prioritize the work, and determine what is ready to release. AI agents accelerate implementation by helping plan, write code, integrate services, diagnose issues, test behavior, and prepare deployments.
The value comes from the operating model—not from handing over judgment. I remain accountable for the decisions, quality, security, voice, and final outcome while using agents to expand the speed and range of what I can deliver.
Product capabilities
Build the tools behind the experience
A private Notes Manager connects to Threads, synchronizes posts and media, and turns a third-party feed into a searchable, paginated experience on my own domain.
The system reads the original post and conversation context, generates reply drafts in my voice, supports editing or emoji-only responses when appropriate, and publishes only after explicit approval.
Accountability
Clear roles for human judgment and AI acceleration
- Vision and use-case selection
- Information architecture and UX direction
- Requirements, priorities, and acceptance criteria
- Content, voice, privacy, and publishing decisions
- Review, quality control, and release approval
- Planning and implementation support
- Code generation and refactoring
- API integration and debugging
- Responsive and functional testing
- Documentation and deployment support
The impact
A broader delivery capability—not just a faster website build
The result is a platform I can operate, extend, and improve continuously. It has shortened the path from idea to working software while allowing me to work fluently across product strategy, experience design, content, data, APIs, automation, testing, SEO, and release management.
More importantly, it demonstrates how I approach AI adoption in practice: begin with a real workflow, design the surrounding system, keep people in control, learn from production use, and expand only where the experience becomes meaningfully better.
What I learned
Agentic development changes the shape of product leadership
Specific direction compounds. Clear outcomes, constraints, and acceptance criteria make agent work substantially more useful.
Inspection is part of creation. Testing the live experience, tracing failures, and refining edge cases are as important as generating the first implementation.
Human review belongs in the workflow. Sensitive actions, public voice, and publishing decisions should remain visible and reversible.
AI fluency is multidisciplinary. The strongest results come from connecting product judgment, design literacy, technical understanding, and operating discipline.