Case study

Demystified AI, Amplified Results

At Baseline, we translate the complexity of artificial intelligence into practical, accessible solutions. Our expertise comes to life through hands-on applications that directly address today’s business challenges.

These case studies highlight how we’ve transformed advanced analytical models into powerful decision-making tools, turned raw data into strategic insights, and automated complex processes to unlock the full human and technological potential of our clients.

Each project demonstrates our commitment: making artificial intelligence not only understandable, but truly usable for your organization.

fond

Librairie Martin

Case Study: Baseline × Librairie Martin

AI Supporting Quebec Bookstores

In collaboration with SaaSpasse

Our Client

Librairie Martin, founded in Joliette, is a well-known name in book retail across Quebec. Faced with the challenges of managing a massive inventory (over 30,000 titles per branch) and difficulties recruiting specialized staff in remote areas, the company was seeking an innovative solution to improve customer service.

Slim, their management software created by booksellers for booksellers, centralizes inventory, ordering, and customer loyalty management. However, one key element was missing to address the current market realities.

The Challenge
Librairie Martin

  • Overwhelming volume: 36,000 new books published each year in the Francophonie
  • Rapid turnover: Between 5,000 and 7,000 new books to integrate each week
  • Scarce expertise: Difficulty finding experienced booksellers, especially in rural areas
  • High expectations: Customers expecting personalized and precise advice

 

How can we empower staff, even those with limited expertise, to provide relevant and high-quality recommendations from such a vast inventory?

 

Our Approach

At Baseline, we turn AI promises into tangible results. For Librairie Martin, our process unfolded over several key phases:

  1. Exploration and Ideation Phase
    • Brainstorming sessions to identify AI integration opportunities
    • Validation of the technical and economic feasibility of proposed solutions
    • Definition of measurable objectives and a clear roadmap
  2. Design Phase
    • Development of a "virtual bookseller" based on a Retrieval Augmented Generation (RAG) architecture
    • Integration with the existing inventory system (Slim)
    • Design of an intuitive interface for employees
  3. Testing and Validation Phase
    • Deployment in a first branch for real-world testing
    • Adjustments based on field feedback
    • Optimization of performance and recommendation accuracy
  4. Implementation Phase
    • Staff training on how to use the new tool
    • Integration with RFID technology for a seamless experience
    • Setup of maintenance and continuous improvement mechanisms

The Solution: An Intelligent Virtual Bookseller

The system we developed allows Librairie Martin employees to:

  • Instantly access relevant recommendations based on customer queries
  • Quickly identify books in stock that match the expressed needs
  • Avoid frustration by only suggesting titles physically available in-store
  • Offer expert-level service without needing to know the entire catalogue

The major innovation: the system only suggests books that are physically in stock, eliminating customer disappointment and maximizing immediate sales.

Tangible Results

"Thanks to the integration of AI, I’ve been able to sell books I probably wouldn’t have sold otherwise, and I can prove it beyond a doubt." - Luc Ayotte, Librairie Martin

Some concrete examples demonstrate the impact of our solution:

  • Increased sales: Books that would have stayed on the shelves are now being sold
  • Continued service: On Father’s Day, without the usual expert staff member, the system enabled the sale of three specialized books
  • Improved customer experience: Clients receive relevant recommendations even when staff are unfamiliar with certain titles
  • Resource optimization: Less reliance on experts for daily recommendations

Lessons Learned

This collaboration strengthened several of our convictions:

  1. AI is not an end but a means – It must solve a real, measurable problem
  2. Humans remain central – Technology enhances staff capabilities without replacing them
  3. Preparation is key – Taking the time to understand needs before development is essential
  4. Flexibility pays off – Prioritizing features that bring immediate value is crucial

Client Testimonial

"When David from Baseline approached me, his first question was: ‘Do you know much about artificial intelligence?’ It was a topic we kept hearing about, but I’d never really explored its potential. [...] At the beginning, we brainstormed several ideas, and each time, David came back with a clear answer: ‘We can do it.’ That confidence and ability to always find a solution convinced me he was the right person to work with."
— Luc Ayotte, Librairie Martin

Next Steps

The collaboration between Baseline and Librairie Martin continues with:

  • Expansion of deployment to other branches
  • API development to connect more bookstores to the system
  • Continuous optimization of the recommendation algorithms
  • Support in scaling up commercial deployment

Our Partnership with SaaSpasse
SaaSpasse logo

This case study was originally featured on the website of our partner SaaSpasse,

Quebec’s go-to resource for SaaS and the local tech ecosystem.

SaaSpasse and Baseline share the same vision: highlighting tech innovations that generate real results for local businesses. Our collaboration allows us to document and share best practices for AI integration in Quebec SMEs.

We thank the SaaSpasse team for their support and for helping promote Quebec’s tech success stories.

About Baseline

Founded in Quebec City in 2020, Baseline is a cooperative specializing in artificial intelligence that transforms AI promises into concrete results for SMEs. Our team supports companies at every stage of their digital transformation.

Our philosophy: We don’t start with the technology to find a problem — we start with your problem and find the right technology to solve it.

Facing a similar challenge?

AI isn’t just a buzzword. In the right hands, it becomes a real growth lever. Contact us to discuss how we can turn your challenges into opportunities.

Book a meeting with our team

Read the original article of SaaSpasse

 

 

 

Learn more

Workleap

Case Study: Baseline × Workleap

Claude-a-thon Workleap: training 350 people on AI, from virtual to in-person

How Baseline designed and delivered the upskilling journey that made a company-wide AI hackathon possible.

Client Workleap
Sector HR SaaS / Productivity
Mandate Pre-event training and on-site AI coaching
Date April 2026 | Montréal

The contextBaseline Team at Workleap

Workleap wanted to do something concrete with AI. Not a conference. Not a theoretical workshop. A real moment where their 350 employees — technical or not — would get their hands on it.

The central question: "How can we optimize our operation and increase our velocity with AI?"

As their CTO put it: "We wanted to do better than classic innovation workshops where you come up with a lot of good ideas, but nothing concrete."

To get there, a real problem had to be solved: teams had very different levels of AI maturity. People who had never opened Claude. Developers who wanted to deploy advanced applications. And everyone in the same room, on the same day.

Sending 350 people into a hackathon without preparation is wasted potential. Baseline was brought in to prevent exactly that.

What Baseline delivered

1. Virtual training → Mastering Claude Cowork

Ahead of the event, Baseline led a comprehensive training on Claude Cowork, designed to make the platform immediately operational for any profile.

On the agenda: understanding the structure of projects and conversations, building effective system instructions, creating and sharing reusable Skills with colleagues, and learning how to guide Claude toward precise results rather than generic answers.

The goal wasn't to discover the tool. It was to arrive on the day knowing exactly what to do with it.

2. Virtual training → Taking action with Claude Code

The second training targeted teams ready to go further: using Claude Code to build real tools, automate workflows, and deliver functional prototypes without waiting for a development team.

Content covered: structuring a development project with Claude, using context strategically, debugging effectively, and understanding how Claude Code interacts with files, terminals, and real work environments.

For non-developer profiles, it's a gateway to technical autonomy. For developers, it's a speed multiplier.

3. On-site AI coaching

On the day of the event, 6 AI coaches from Baseline and Gaiia supported 55 mixed teams of 5 to 7 people during 3.5 hours of effective build time.

The directive was clear: guide without putting your hands on the keyboard. Unblock, don't direct.

Between the two build sessions, coaches synced up to identify which teams needed a boost in the afternoon. A simple mechanic that kept teams from drifting toward slide-polishing instead of actual building.

Non-technical teams were the most active: how to access transcripts to automate a workflow, how to extract and monitor client information to spot opportunities in real time. Concrete business problems, not classroom exercises.

On the tech teams' side, conversations touched on more advanced concepts: modules capable of automatically listening to external events (Webhook-type), for example, triggering an action in Claude as soon as a new employee is added to an HR system, for smarter onboarding tools.

The numbers

2 Virtual training sessions led by Baseline
6 On-site AI coaches (Baseline + Gaiia)
55 Teams supported during 3.5 hours of effective build time
350 Employees ranging from "never opened Claude" to "advanced developer"

What the training made possible

"I'm still in shock."
— Guillaume Roy, co-CEO of Workleap

Teams from every level and every function delivered functional projects in a single day. People who said they were "not comfortable with AI" left with something demo-able. That's not a coincidence: it's the result of preparation done right.

For the details of what was built and measured during the event, see Workleap's article in Training Industry.

What we take away

A Claude-a-thon without upfront training is wasted potential.

  • AI is better learned by doing than by listening. But "doing" without preparation produces chaos.
  • Baseline's value: train before, unblock during.

One question remains open for what comes next: how do you make the prototypes built accessible to the whole organization, securely, without flooding the Product team? The next step is to map out a clear path where deployment fits within the security practices and regulations already in place. We're closely watching what Anthropic might deploy to bridge that gap.

What's next

For teams that got a taste of speed, some will want to go further. That's where ongoing support makes all the difference: turning a prototype into real practice, keeping the momentum going, and continuing to build skills without waiting for the next event.

Are you planning an AI activation event for your organization?

Beyond custom AI development, Baseline designs learning experiences that produce measurable results, not inspiring days that evaporate by Monday morning.

Let's talk about what that could look like for you

 

 

 

Learn more
This website uses cookies to improve your browsing experience.
This includes essential cookies required for the site's operation, as well as cookies used for anonymous statistics, ensuring a smooth user experience, and displaying personalized content. You can approve the desired categories. Please note that depending on your settings, some site features may be disabled.
Privacy policy
This website uses cookies to improve your browsing experience.
This includes essential cookies required for the site's operation, as well as cookies used for anonymous statistics, ensuring a smooth user experience, and displaying personalized content. You can approve the desired categories. Please note that depending on your settings, some site features may be disabled.
Privacy policy
Your cookie preferences have been saved.