AI Product Platform / Automated Commerce Workflow
FYKZIT started with a simple product idea: instead of making people search through hundreds of products themselves, let them describe the problem they're trying to solve and build the product discovery workflow around that problem.
I designed the product experience and the automated system connecting user input, AI decision logic, external product data and the final interface.
THE PROBLEM
Traditional product discovery starts with knowing what you want to buy. But users often know the problem, not the product. Someone might know what is uncomfortable, inconvenient or difficult in their daily life without knowing which combination of products could actually help. FYKZIT reverses the normal shopping flow. Instead of: Product → Search → Compare the experience begins with: Problem → Understand → Solution → Products That required more than designing a search interface. The system needed to interpret an open-ended problem, turn it into something actionable and connect that decision to actual product discovery.
THE PRODUCT IDEA
The core FYKZIT experience begins with one input:
“What problem are you trying to solve?”
The user's input enters an AI-driven decision workflow. Rather than immediately returning random products, the system first works on understanding the problem and narrowing the direction. The interface then presents possible paths so the user can choose the solution that best matches what they actually need. That choice becomes the trigger for the next part of the system.
FROM INPUT TO AUTOMATED ACTION
The important part of FYKZIT happens behind the interface. User input needs to move through several stages before the final product bundle can appear. The workflow was designed around:
User Problem → AI Logic → User Choice → Product Search → Filtering & Ranking → Bundle → Interface
Different parts of the process communicate through automated scenarios, webhooks and structured data. Instead of manually selecting products for every possible problem, the system can turn the selected solution path into a product-search operation. This is what transformed FYKZIT from a designed interface into a working automated product workflow.
AI DECISION LOGIC
AI is used to understand the user's input and turn an ambiguous problem into structured directions that the application can work with. The goal wasn't to add a chatbot. The AI layer exists because the next part of the product requires structured information. The system takes natural-language input and converts it into actionable output that can trigger the next workflow. This keeps the AI role specific: understand the problem → structure the decision → move the application forward.
CONNECTING THE WORKFLOW TO REAL PRODUCTS
Once a direction is selected, the workflow moves from interpretation to product discovery. I worked on connecting the application to external product data so product recommendations could be generated dynamically rather than stored as static recommendations for every possible query. Product information returned from the external source is then prepared for the application interface. This meant coordinating AI output, automated scenarios, API responses and frontend data rather than treating them as independent pieces.
PRODUCT FILTERING & RANKING
Returning products isn't enough. The system also needs to decide which products are worth showing. I developed ranking logic around signals such as: Relevance — whether the product actually matches its intended role Rating — stronger ratings are preferred Orders — products with more completed orders are favored Price — avoiding suspiciously cheap or unnecessarily expensive options Images — prioritizing listings with usable product presentation The goal was to move from: “The API found something.” to: “The system selected something worth presenting.”
THE BUNDLE
FYKZIT was designed around a bundle rather than one isolated recommendation. The selected solution can return a structured set of products that work together around the user's problem. The bundle is then displayed through the product interface instead of exposing the underlying API response to the user. This separation was important. The backend workflow can change how products are found and ranked without forcing the user experience to become a technical search interface.
REUSABLE PATHS INSTEAD OF REPEATING WORK
The product architecture also considered what should happen when a useful solution already exists. Rather than treating every user input as a completely new request, the workflow was designed around reusable solution paths and product bundles where possible, while allowing missing or weak product results to be filled dynamically. For new problems, the system can create a temporary path, perform product discovery and prepare a result that can later be reused. This turns repeated queries into product knowledge instead of repeatedly starting the entire process from zero.
THE SYSTEM BEHIND THE INTERFACE
FYKZIT combines several layers of product work: Product and UX design Web application interface AI-driven decision logic Automated workflows Webhooks External product API integration Structured data mapping Filtering and ranking logic Dynamic product bundles The value isn't any one of those technologies individually. It's the workflow connecting them.
THE RESULT
FYKZIT became a practical example of how I approach product development beyond the interface. I started with a user problem, designed the interaction around it and then worked backward into the logic and integrations required to make that interaction actually function. The result is not simply an AI interface or a product search page. It's a connected workflow where a natural-language problem can trigger decision logic, automation, external data processing and a structured result inside the product experience.