Inspiration

Small businesses power local communities, yet many still make critical merchandising decisions without understanding how customers actually move through their stores. Sales data only shows the outcome — not the journey. We were inspired by the invisible effort behind every small shop: the late nights, carefully arranged shelves, handwritten signs, and constant experimentation that often goes unseen. We wanted to build something that gives store owners clarity instead of guesswork. Pathwise was created to help businesses visualize customer behavior, understand visibility, and make smarter in-store decisions backed by data.

What it does

Pathwise is an AI-powered retail intelligence platform that helps businesses understand shopper behavior inside physical stores. Users can upload floor plans, product placement data, and sales information to generate actionable insights about visibility, customer traffic, and store performance.

The platform:

• Simulates customer movement and attention patterns
• Generates heatmaps showing high and low visibility zones
• Predicts product performance based on placement
• Suggests optimization strategies for shelves and layouts
• Identifies high-performing product adjacency pairs
• Allows side-by-side comparison of store layouts
• Provides AI-generated retail insights and learning reports

By turning physical store data into visual intelligence, Pathwise helps businesses improve visibility, increase engagement, and optimize sales.

How we built it

We built Pathwise using a full-stack architecture focused on scalability, simulation, and AI-powered analysis.

Frontend

• Next.js
• React
• TypeScript
• Tailwind CSS
• Framer Motion
• D3.js / Recharts

Backend

• Node.js
• Express.js
• FastAPI microservices

AI & Simulation

• OpenAI APIs
• LangChain
• PyTorch
• Scikit-learn
• NetworkX graph simulations

Data & Infrastructure

• PostgreSQL
• Prisma ORM
• Redis
• AWS S3
• Docker
• Vercel

We designed custom behavioral simulation systems that model different shopper types — goal-oriented shoppers versus exploratory shoppers — and used pathfinding algorithms to visualize movement, attention, and visibility throughout a store.

Challenges we ran into

One of our biggest challenges was translating real-world retail behavior into a digital simulation. Human movement inside stores is unpredictable and influenced by countless variables, making it difficult to model accurately.

Another major challenge was connecting multiple forms of data — floor plans, product locations, sales records, and traffic simulations — into one unified system that remained understandable for non-technical users.

We also spent significant time balancing technical complexity with usability. While our backend simulations and AI systems were advanced, the platform still needed to feel simple, intuitive, and actionable for small business owners.

Accomplishments that we're proud of

We are proud of building an end-to-end platform that transforms invisible customer behavior into understandable insights for small businesses.

Some accomplishments include:

• Creating a working customer movement simulation engine
• Building interactive visibility heatmaps
• Designing AI-generated optimization recommendations
• Developing comparative layout analysis tools
• Successfully integrating machine learning and LLM reasoning into retail workflows
• Creating an interface accessible to non-technical users

Most importantly, we are proud that Pathwise focuses on empowering small businesses with tools typically reserved for large enterprise retailers.

What we learned

Through building Pathwise, we learned how complex human behavior can be when translated into systems and algorithms. We gained experience in simulation design, AI orchestration, data visualization, and full-stack system architecture.

We also learned the importance of designing technology around people rather than just features. Small business owners do not need more raw data — they need clarity, context, and actionable insights.

Additionally, we learned how combining machine learning with thoughtful UX design can make highly technical systems feel approachable and useful.

What's next for Pathwise

Our next goal is expanding Pathwise into a more intelligent and scalable retail optimization platform.

Future plans include:

• Extending prediction windows from 30 days to 90+ days
• Incorporating seasonal and weather-based retail forecasting
• Adding real-time customer tracking integrations
• Supporting multi-location business analytics
• Introducing mobile dashboards and notifications
• Enhancing behavioral AI models with richer shopper personas
• Building collaborative planning tools for retail teams

Ultimately, we want Pathwise to become the operating intelligence layer for physical retail spaces — helping businesses understand not just what customers buy, but how they think, move, and interact inside stores.

Built With

• python