.svg "Baseline-logo-complet-blanc (9)"){kind=logo}
Retail | AI Solution Development - RAG
.png?width=1500&height=450&name=Banni%C3%A8re%20Baseline%20(14).png "Bannière Baseline (14)")
Client: Tanguay
Year of Completion: 2024-2025
[Context and Challenge]
Tanguay is one of the leaders in the furniture industry in Quebec. A significant portion of their sales occurs online, particularly outside business hours, without customers having access to sales associate advice. This reduces the opportunity to suggest complementary products and enhance the customer experience. Tanguay wanted to improve customer experience and increase online sales outside business hours by offering personalized recommendations without human intervention. Tanguay's website receives approximately 336,000 monthly visits. The agent currently supports about 10% of traffic.
[Approach]
We implemented a Retrieval Augmented Generation (RAG) solution, using a Language Learning Management System (LLM) (e.g., ChatGPT) to interpret customer needs and suggest products based on their queries, leveraging a vector database (Weaviate). Specific cross-selling rules and tailored product descriptions were integrated to provide personalized recommendations. A semi-automated evaluation system ensures the quality of the responses generated by the chatbot.
[Results]
A fully deployed AI conversational solution used in real-world context, capable of interpreting complex queries and suggesting relevant alternatives. For example, when faced with an atypical request for an "8-door fridge," the chatbot understands the intent behind the query and suggests 4-door models, demonstrating its ability to grasp the customer's actual need and recommend coherent options.
Integrated into the service environment, the solution acts as a true virtual advisor, accompanying customers throughout their purchasing journey and facilitating access to product information.
See the chatbot in action on Tanguay's website.
Retail | AI Solution Development - RAG
Client: Librairie Martin
Year of Completion: 2024
[Context and problem]
Librairie Martin was struggling to recruit clerks capable of providing quality book recommendations, which was negatively impacting customer experience and sales. Maintaining a personalized service, aligned with customer preferences and in-store inventory, became essential. Beyond customers who physically visit the bookstore, this bookseller receives 8,100 visits to its website monthly. The agent currently handles approximately 50% of this traffic.
[Approach]
We developed a Retrieval Augmented Generation (RAG) solution that performs searches in a vector database, which is updated daily with the bookstore's inventory, and returns responses generated by a large language model based on artificial intelligence. The solution also integrates transcription and speech synthesis technologies that allow users to interact with the system using voice commands.
[Results]
Librairie Martin has an intuitive recommendation system integrated with its inventory management system. The chatbot can answer questions about the entire catalog, allowing staff to focus on niche genres like manga. Both customers and staff appreciate the chatbot's recommendations, and its use has enabled the bookstore to serve a larger customer base.
Technology | Development of a multi-label recommendation/classification system
.png?width=1500&height=450&name=Banni%C3%A8re%20Baseline%20(12).png "Bannière Baseline (12)")
Client: Brix Labs
Year of Completion: 2020
[Context and problem]
In this era of innovation and progress, the artificial intelligence revolution is transforming industries. Brix Labs is developing a digital productivity platform dedicated to the publishing world. The company aims to provide publishers with a solution that allows them to automatically assign BISAC codes, an international standard for book classification, to a book. For example, the book "The Billionaire’s Vinegar" holds the codes "Antiques & Collectibles - Wine" and "True Crime - Con Artists, Hoaxes & Deceptions", while the book "Night Watch" displays the codes "Fiction - Literary", "Fiction - Medical", and "Fiction - Civil War Era". In total, there are over 5,000 codes, and a single book can be assigned up to three different codes.
Initially, the assignment of BISAC codes is done manually by the publisher, a particularly tedious task. The complexity of this process also results in the underutilization of certain BISAC codes, which can hinder the marketing of the books.
[Approach]
Baseline developed a recommendation system that extracts the five most relevant BISAC codes based on the book's summary. The structure of the summary and the words used provide enough information to accurately characterize the book. The infrastructure surrounding the system allows it to continuously improve and suggest higher-quality recommendations.
[Results]
The system's recommendations are reliable 90% of the time. The remaining 10% primarily comes from BISAC codes that the publisher did not use. The prediction occurs in less than a second, enabling Brix Labs to "call" the system in real time and integrate it with their technological tools.
Technology | Multi-agent architecture
-1.png?width=1500&height=450&name=Banni%C3%A8re%20Baseline%20(12)-1.png "Bannière Baseline (12)-1")
Client: Serko
Year of Completion: 2026
[Context and problem]
Serko, a technology company specializing in business travel management, wanted to completely transform the user experience of its platform. The goal: to move from a traditional interface to an Agentic Experience (AX), where users express their needs in natural language and AI automatically manages the search and recommendation of personalized flights and hotels, taking into account their preferences, history, and company policies.
[Approach]
Baseline designed a multi-agent architecture powered by PydanticAI, capable of handling complex requests and providing personalized responses. To effectively manage personalization, we developed a hybrid system combining user preference learning and rapid filtering of results. A short- and long-term preference model allows recommendations to be adapted to travelers' overall habits while taking into account the specific needs of each trip.
[Results]
The proof of concept quickly demonstrated its potential: even before its final delivery, Serko decided to accelerate the development of the beta version and extend its collaboration with Baseline. This project positions Serko at the forefront of intelligent conversational experiences in the business travel sector.
Health | Conversational AI
-1.png?width=1500&height=450&name=Banni%C3%A8re%20Baseline%20(13)-1.png "Bannière Baseline (13)-1")
Client: BSW Health
Year of Completion: 2026
[Context and problem]
Baylor Scott & White Health (BSW Health), the largest non-profit healthcare system in Texas, wanted to simplify the use of its complex and information-heavy digital platform, myBWSH. The goal was to create an agentic experience that would allow users to easily search for information and perform transactions, schedule appointments, pay bills, and manage their coverage through a centralized, chat-based entry point. While BSW Health had developed an internal proof of concept, the team wanted to validate the experience with specialists and accelerate development to achieve its objectives.
[Approach]
Baseline is supporting BSW Health in validating and optimizing their agentic concept. We analyze user journeys, test the feasibility of conversational interactions, and propose adjustments to simplify navigation while maintaining the complexity of essential functionalities. Our role is to transform the proof of concept into a robust prototype, ready to move to the accelerated development phase, while ensuring a smooth, intuitive, and user-centric experience.
[Results]
The collaboration with Baseline will allow BSW Health to validate and optimize its proof of concept, accelerating development towards a more robust and efficient version. The solution aims to offer users a more intuitive, personalized, and centralized experience, positioning BSW Health as an innovative player in digital health services.
Why choose Baseline
Cutting-edge expertise in AI
End-to-end support
Concrete and measurable results
Our AI development method
Developing a high-performing and sustainable AI solution is not something that can be done on the fly. At Baseline, we follow a structured and agile approach that puts you at the heart of the process.
-1.png?width=850&height=1500&name=Banni%C3%A8re%20Baseline%20(7)-1.png "Bannière Baseline (7)-1")
Innovation canvas
Co-construction of the business project to align the technological vision of AI with the strategic objectives of the company.
Statement of Work
Concrete definition of requirements (functional and non-functional), constraints (technical, regulatory) and validation of the sufficiency and accessibility of the data.
Development plan
Breaking down the overall vision into development stages to balance risk management and continuous value delivery.
Prototyping
Rapid validation of the AI-based technical approach, adjustment of the trajectory and construction of a usable prototype serving as the foundation for the final system.
Application development
Building the remaining functionalities around the prototype, using proven iterative methods, and integrating with existing systems (ERP, CRM, etc.).
Deployment
Making functionalities available at scale, solidifying the infrastructure and setting up monitoring and observability tools to detect failures.
Training and knowledge transfer
Training of users and internal teams for the use and maintenance of the solution, ensuring autonomy through clear documentation.
Maintenance évolutive
Post-deployment support to ensure stability, security, adaptability, and to provide updates and integration of new features as needed.