After more than 15 years working in software development, mainly in ecommerce, I began exploring artificial intelligence in greater depth.

During my studies, experiments, and development of AI systems, one opportunity caught my attention: the potential of AI shopping assistants to transform how consumers interact with online stores.

The idea seemed simple: intelligent, personalized assistance available 24 hours a day. As I explored it further, I realized that building something useful for ecommerce required much more than connecting an AI model to a store's catalog.

That is how Foloon began.

What is Foloon?

Foloon is an AI Shopping Assistant developed specifically for ecommerce. Its goal is to help consumers throughout the shopping journey, from product discovery to after-sales support, with a personalized experience informed by context.

Assistance can be reactive, when consumers ask for help, or proactive, when the assistant identifies a relevant moment to offer guidance, recommendations, or support.

For merchants, Foloon offers tools to configure AI behavior, personalize communication, manage knowledge and commercial strategies, and monitor consumer interactions.

From the beginning, Foloon was designed to work across different ecommerce platforms. It combines artificial intelligence, consumer assistance, commercial strategies, merchant control, and an architecture designed to grow.

The problems I identified

For consumers

Traditional online shopping still relies heavily on search, filters, categories, and product pages. These tools require consumers to know how to find what they need.

Shoppers often have to compare products, interpret technical specifications, understand differences between models, or look for information about availability, delivery, and returns. An experienced salesperson can help with these decisions in a physical store, but that assistance is not always available online.

An AI shopping assistant can help consumers find products, compare alternatives, clarify questions, and make decisions with greater confidence.

Consumers do not always know they need help or that a tool exists to guide them. I believe proactive interaction will play an important role: offering help at relevant moments without requiring the consumer to start a conversation, interrupting browsing, or getting in the way of checkout.

For merchants

Each store has its own catalog, brand identity, commercial policies, support practices, and sales strategies. Merchants need control over how their assistant behaves, which information it uses, how it communicates, and how it participates in the shopping experience.

They also need to define interaction strategies, recommendations, cross-selling, upselling, and other actions suited to their business.

Conversations can reveal recurring questions, product discovery difficulties, consumer needs, and opportunities for improvement. That information needs to be accessible and useful to the people running the store. This combination of needs shaped Foloon.

Beyond a chatbot

For years, stores have used chatbots to answer frequently asked questions, automate support steps, or route customers to a human team. Those functions remain useful, but AI can also participate directly in the shopping experience.

An AI shopping assistant can understand conversational context, interpret needs, recommend products, compare options, and guide decisions. It can also offer proactive assistance similar to that of an experienced salesperson.

That assistance must respect the consumer's situation and contribute to the purchase decision without becoming intrusive. Helping the market understand the value of this experience is one of the challenges of adoption.

An architecture for multiple platforms

One of my main decisions was to avoid limiting Foloon to a single ecommerce platform. Each platform has its own APIs, data structures, authentication mechanisms, catalogs, and operational characteristics.

I developed a modular architecture with a central core responsible for AI and the system's main rules and features. Platform-specific integrations are separate from that core.

This lets Foloon adapt to different ecosystems without rebuilding the entire solution, while centralizing maintenance, updates, agent improvements, and performance optimization.

The vision is to reach as many stores as possible, regardless of the technology they use.

What Foloon offers

Consumer assistance

The assistant was developed to support different needs throughout the shopping journey:

  • Product search and discovery.
  • Answers to product questions.
  • Personalized recommendations.
  • Product comparisons and similar alternatives.
  • Guided shopping.
  • Order information, status, and tracking.
  • Answers about shipping, returns, refunds, and store policies.
  • Conversation continuity and session memory.
  • Handover to human support.
  • Related product recommendations, upselling, and cross-selling.

Proactive interaction

Foloon's proactive interaction is designed to offer help at relevant moments, considering the context and the merchant's strategies. This can include recommendations, related products, and guidance on alternatives.

The guiding principle is that AI should help consumers and respect the shopping experience. The evolution of these features includes control over when, how, and why an interaction takes place, with the aim of making each interaction more relevant.

Foloon AI Studio

AI Studio lets merchants configure their agents' behavior, including:

  • Brand voice and communication style.
  • Knowledge bases and custom content.
  • Search and recommendation rules.
  • Custom instructions and intents.
  • Integrations and operational settings.
  • Specific assistant behaviors.

This allows each assistant to reflect the characteristics of its store.

Foloon Sales

Foloon Sales represents the evolution of intelligent assistance toward a more active role in the merchant's commercial strategy.

Planned capabilities include commercial behavior settings, proactive interaction strategies, related product recommendations, cross-selling, upselling, discount strategies, and recovery and assistance actions throughout the shopping journey.

The goal is to let merchants define their own strategies and retain control of the assistant. Foloon Sales supports different commercial configurations on a common architecture. These capabilities are part of the solution's evolution and are being expanded progressively.

Foloon Analytics

Foloon Analytics brings together information about AI activity, usage, messages, identified intents, and other operational indicators.

These insights help merchants monitor the solution and understand consumer interactions. The vision is to keep developing this analytical capability so conversations can inform commercial decisions and improvements to the shopping experience.

Cloud and On-Premises

Foloon Cloud

The Cloud version follows a SaaS model, with infrastructure and operations managed by Foloon. It brings together the AI Shopping Assistant, AI Studio, analytics, management, and personalization tools in a centralized environment.

The goal is to simplify adoption so merchants can configure and use their assistants without managing all the technical infrastructure. Foloon Cloud is still in beta and has not been launched.

Foloon On-Premises

The On-Premises version was developed for companies that prefer to run the solution on their own infrastructure. It gives merchants greater control over the environment, data, and application code.

The architecture can accommodate each company's infrastructure and integration needs while preserving the core assistance features. The On-Premises solution is ready.

These two approaches reflect the different technical, operational, and commercial needs of different businesses.

The challenge of building Foloon

Foloon has been one of the broadest projects of my career. As a solo founder, I was responsible for practically every stage: the brand, name, visual identity, logo, interface design, frontend and backend development, ecommerce integrations, AI agent architecture, processing services, management systems, and infrastructure.

I also developed and configured the AWS infrastructure, including workers, caching, databases, private networks, security controls, CI/CD, deployment automation, and testing strategies.

Balancing intelligence, performance, and cost

Understanding requests and producing useful answers is only part of the challenge. An assistant operating in a real store also needs to respond quickly and remain affordable.

During development, some requests took approximately 14 seconds to process. I revisited the agents' processing architecture, task organization, prompt strategies, and model usage.

By combining processing approaches, reducing unnecessary operations, and using reasoning selectively according to each request's complexity, simple product queries came to be processed in approximately 200 to 500 milliseconds.

The challenge was balancing intelligence, answer quality, speed, and operating costs. Using advanced models requires an architecture that employs them efficiently. This work was fundamental to making Foloon viable for stores of different sizes.

Building a complete system

The assistant needs to work with product and order information, commercial policies, individual store settings, and multiple services. The system also needs security, data isolation, performance, scalability, and ongoing maintainability.

Integrating these responsibilities required architectural decisions that went beyond implementing individual features. My goal was to develop a complete product with a technological foundation that could evolve.

My vision for the future of AI shopping assistants

I believe AI shopping assistants will become an essential part of ecommerce, following a path similar to search, payment systems, and management tools. Every store needs a way to help consumers at any time.

I see assistants understanding consumer needs, offering relevant recommendations, participating in commercial strategies, and interacting proactively throughout the shopping journey.

Despite the growth of this market, I have not yet identified a major independent player established as its definitive reference. I see an opportunity for specialized companies whose core product is this experience.

Adoption will depend on demonstrating results, making solutions accessible, simplifying integrations, and giving merchants confidence and control over the AI representing their businesses.

My vision for Foloon is to contribute to this transformation with a specialized, flexible solution that can evolve with ecommerce.

Final thoughts

Foloon began as an opportunity I identified while studying and experimenting with AI. It evolved into a system combining AI engineering, modular architecture, ecommerce integrations, management tools, commercial strategies, and different operating models.

Developing it independently required applying more than 15 years of software engineering experience while learning and experimenting with new approaches.

There is still considerable room to grow. The original vision remains: making intelligent assistance a natural part of online shopping, combining support, recommendations, proactive interaction, and commercial strategies to benefit consumers and merchants.

Foloon represents my contribution to that future.

Because answers can't wait.

Explore Foloon