Our Investments

Kruger aims to improve the fiber strength consistency by developing new approaches to reduce the negative impact on inconsistent recycled fiber quality caused by the variability of the fiber at the beginning of the process and understand how we can act effectively to optimize output. This will be followed by better predicting the output quality of the fiber to minimize production cost. This project will focus on implementing advanced predictive & prescriptive analytics supported by Machine Learning (ML), to predict and calibrate fiber quality to provide better quality control.

The project consists of developing optimizers to detect visual and geometric defects by using deep learning on images from our production lines for first and second raw wood transformation.

The scope of this project is to develop a machine learning driven electronic media detection algorithm. This detection solution will be integrated with our existing x-ray scanner technology for customers that are seeking automated detection assistance in order to protect ingress and egress of data bearing devices into / out of their facilities, which could pose a risk to data security. By automatically identifying and alerting the scanner operator of the potential presence of these devices, the tool reduces the dependence upon visual detection. This task can be incredibly challenging, particularly for smaller media devices (e.g. SD card, USB sticks, etc.).

The project consists of upgrading and automating our REC sorting center with intelligent robots to allow for greater operational flexibility and an increase in the volume, capture rate and quality of outgoing materials.

Compro, an innovative company in the field of voice messaging, wants to automate the process of writing voice message scripts. Largely thanks to natural language processing, Compro wants to automate the entire process, from the collection of production ideas to the automatic generation of scripts. This automation will reduce the pressure on editors and will allow them to focus on improving the written texts.

The aim of the project is to use alternative data to forecast facilities usage inside a public infrastructure to adapt maintenance scheduling in order to improve service levels while using less resources.

The project aims to gain efficiency through autonomous sourcing. Concretely, the algorithms developed will identify and then analyze thousands of CVs contained in databases and to rank candidates for a given job description using AI.

Integration of deep learning in the automated sorting of a recyclable material stream to optimize the process.

Promoting innovation as a business model, Boa-Franc wishes to implement artificial intelligence in its processes to optimize them. Boa-Franc’s first phase of the project is aimed at the dynamic scheduling of finishing operations. The second part of the project will focus on the automation of dryers, which will mainly deal with the automation of drying decisions and the dynamic scheduling of drying activities.

Adaptation and deployment of an industrial grade perception and cognition platform allowing the use of industrial robots to apply complex surface treatment processes without requiring manual programming, off-line programming or even the use of part location templates. The artificial intelligence platform requires the modeling of each process, the creation of a virtual twin of each robotic cell and the modeling of the application constraints of each process. Finally, the objectives to be achieved for each part must also be defined. Automatic optimization by AI should exceed the performance of a human generated program.

Development and deployment of AI-Powered solution to optimize Cascades’ logistic network. AI algorithms will evaluate millions of routes lasting from 1 to 5 days. The tool is going to find the optimal routes that minimize empty mileage while respecting business and legislation constraints.

The Osiris project consists of a series of insights generated via predictive models to dynamically estimate the selling price by analyzing several complex variables so that the Krōps Group’s sales departments can know the trends in the Canadian fresh fruit and vegetable market, as opposed to the current framework based on personal interpretation of available informal data. Osiris will be a revolution in the industry and will enable Krōps teams to generate value with suppliers, customers as well as the company itself.

The ‘lift model’ that Wavo is developing is a machine learning model that, given a variety of data (social time-series metrics, music industry metrics, artist and track metrics), aims to quantify the incremental effect of Wavo’s advertising. We are building a lift model specific to the music industry that predicts the impact that marketing and advertising provide above native demand – ie, the increase in sales, listeners and other key performance indicators (KPIs) gained that would not have occurred without marketing efforts.

The project is for the development, deployment and enhancement of artificial intelligence (AI) assisted screening for vascular changes in the retina using retinal fundus photography and Optical Coherence Tomography (OCT). The use of these diagnostic instruments and software will allow optometry clinics to screen for various retinopathies with the goal of early detection. The objective is to build a model with 3 algorithms incorporating the specifics of the Quebec population, the methodology of the IRIS optometrists to grade retina images and the equipment used (Fundus and OCT) to predict, with high accuracy, the presence of a macular irregularity at the point of care, in a pre-test mode with a high level of confidence. For reference, pre-testing is a series of measurements taken by a technician, often using diagnostic equipment. These measurements are then transferred to the optometrist for interpretation prior to, during or after the patient appointment.

The six AI-based initiatives will point to sales, ad placement sales, readership lifecycle study, front page recommendations and automatic news generation to enable quality journalism.

The solution aims to maximize the production of compliant products by optimizing the use of human and machine resources (robots/automats). The connected data will be exploited by predictive and prescriptive models allowing greater flexibility on the temporal layout of the tasks. The solution includes tool life prediction for exotic materials such as Inconel, where tool life usually behaves stochastically. These predictions will then feed the control system.

This project aims to integrate the artificial intelligence into the data process pipeline, to automate the data interpretation process of the non-destructive inspection by visual imaging of large infrastructures, under the frame of the Industry 4.0 revolution.

Patates Dolbec wants to implement artificial intelligence in its potato sorting process. Using a modern vision algorithm based on deep neural networks and the installation of sensors and cameras, we want to detect defects directly on the production line. The sorting of these defects, 30% of which escape the upstream industrial sorting solution (NewTech), is currently done by visual inspection by two workers. In addition to detecting defects, the workers must manually remove the potatoes, a task that is imprecise (30% of rejects are unnecessary) and promotes injuries due to repetitive movements.

The project aims to develop a personal assistant that will allow visually impaired people to recognize objects handled in everyday life, detect objects of interest and converse in natural language with a conversational agent in order to grasp the main characteristics of nearby objects. The personal assistant will be composed of a conversational agent running on the cloud, a software application running on a HumanWare proprietary embedded platform and allowing to recognize/detect objects and to communicate with the conversational agent via web services.

Automation of the product quality assurance function and raw material optimization through the integration of artificial intelligence.

In the first phase, RPM eco wants to use predictive models of recyclables accumulation based on customer history, by market segment, to plan its pickup routes for recycling materials. In a second phase, it wants to use artificial intelligence to replicate the work of dispatchers who assemble routes in order to reduce resource use, maximize volume per truck, reduce unprofitable trips, improve customer satisfaction, and most importantly, reduce greenhouse gases (GHG)

The Ceres project consists of making the purchasing and in-store processing operations for the meat and produce departments more efficient by generating quality demand forecasts to allow the ordering process to be automated. The forecasts generated will improve the labor planning, purchasing, processing and replenishment processes. The project aims to contribute to the strategic objectives by limiting waste, lost sales and gaining operational efficiencies.

We aim to achieve an optimized solicitation system through the implementation of NLP (Natural Language Processing) algorithms and the addition of neural layers (Deep Learning), allowing to substantially increase the probability of matches between a recruiter and a prospect while minimizing human intervention in the laborious and complex recruitment process.

The BIM experts of BIMone firm are tasked with auditing 2D construction plans, ensuring consistency between 3D models and 2D plans and the production of models. To accomplish this job, BIM experts must navigate manually through thousands of unrelated 2D plans on a daily basis. This generates a considerable loss of time and efficiency for our BIM experts and leaves a significant margin of error in the client’s deliverable. The aim of this project is to develop a navigation aid tool for BIM experts in order to make their work efficient and generate a return on the investment.

Novo SST wishes to develop a chatbot / conversational device with an avatar (PAM) and based on a correlation algorithm to process the declarations by linking risks and measures of the prevention plan to allow a recommendation to the Novo SST employee in a first step (Quick Wins project), which will allow to set up in parallel a reinforcement learning model to automate this logic in the Large Scale project

In the context of this program GE Aviation will perform one step towards cognitive robotics in the aviation manufacturing environment. Robotic systems, embedding deep learning and reinforcement learning algorithms will address classification, detection and motion planning challenges posed by the industrial manufacturing setting.

Litière Ouellet produces farm animal bedding, commonly called “shavings”. The objective of this project is to use AI to control the feeding of the bags in order to have a constant volume inside them, limiting human intervention to a minimum and controlling the quality of the shavings. Our operations are now massively dependent on the expertise of our employees and our current machinery. AI will allow us to optimize and automate certain activities to obtain the desired productivity gains.

This project aims to accelerate the Cardiac Magnetic Resonance workflow by developing integrated and automated AI tools for reconstruction, enhancement, quality control, segmentation, and clinical metric extraction of novel ultrafast scanning protocols.

AV&R is active in the aeronautics industry where part inspection is critical to aircraft safety. In order to reduce human intervention, AV&R wants to improve its detection performance, reduce the time to create recipes and automate the validation of recipes by using artificial intelligence.

The project aims to develop a tool to automate the process of analysis and allocation of translation requests. An NLP analysis algorithm, coupled with a series of predictive models, will provide the necessary estimates (nature, complexity, cost, duration) for the allocation of the project to the most appropriate resource.

1- Predict the number of new units rented per month of a potential Senior Residence (RPA) site while respecting the minimum rental price.
2- Prescribe the optimal rental price list to optimize the operation revenues while respecting the rental timeframe.

Implementation of a SCADA system and optimization of the diaper recycling process by continuous artificial intelligence analysis in order to improve production consistency and product quality, increasing by at least 20% the quantity of treated rejects.

The goal of the project is to automate data entry into the company’s ERP system in order to improve efficiency in managing compliance with aerospace standards. As a first phase of this project, we want to automate the capture of invoices, the management of calibration certificates and the management of compliance certificates. We decided to start our business process transformation using AI on these processes because we have a lot of data (over 60,000 invoices) already pre-classified and structured for this project.)

In order to deliver a compelling VR experience, Felix & Paul Studios (FPS) needs to correct the stereoscopy and stitching by adjusting the perspective of the foreground, differently than the background. In the context of the Space Explorers project with NASA with footage captured in the International Space Station (ISS) we must segment foreground (humans and props) from the background of the ISS, especially in the action shots, in stereo and 360 degree footage. This segmentation process is usually performed manually by compositing artists (Rotoscopy). It is very labour intensive, extremely repetitive for the artists and expensive for Felix & Paul.

The ComplisIA program aims to equip Videotron’s Customer Experience Management with an Artificial Intelligence capable of minimizing operational frictions and increasing the desirability of its services. The program supports all activities related to the exploitation of analytical data around the customer in a Big Data environment. It also supports the exploration, operationalization and continuous improvement of self-learning analytical models.

Through the commercial development of Celliers Intelligents, restaurateurs will be able to acquire accurate predictive provisions that will save them the time required to manage the wine inventory of their business.

The objective of the project is to automate the process of medical evaluation and patient triage via a dialogue system (chatbot) while improving medical quality and patient experience. The system would enhance Dialogue’s virtual care service.

The main objective of the project is to improve the discoverability of the content of the Stingray Music, Qello Concerts and Stingray Classica applications, available on Android phones, iPhone and on the web, by providing the user with better adapted content recommendations and by personalizing the content to the user. We want to use artificial intelligence technologies, more specifically deep networks, to build predictive and classification models that will help display content adapted to the situation and interests of users.

The project solves the largest problems for Inovestor: Data collection issues that prevents us to maximize data usage, and offer a broader product selection that will allow for tailored client solutions, by Increasing operational efficiency via information extraction and topic modelling of Comprehensive Annual Financial Reports (CAFR) and Leveraging deep learning techniques to create an AI-powered Factor-based Investing Center to help its customers adopt a more robust and disciplined approach to investing.

Project to develop, adapt and validate a platform for analyzing and learning unstructured documents in the insurance industry, particularly documents from the group insurance sector. With the collaboration of partners/clients from the insurance industry, the project aims to develop a platform that can be used in real-life situations and that can be used worldwide. The project is based on technologies, algorithms and AI models. Eventually, it will allow insurers and intermediaries to analyze group insurance contracts more finely and thus improve their productivity, speed and profitability.

As an early step in drug discovery, hit identification involves testing molecules in a cellular or biochemical model of disease. Molecules that show early signs of efficacy in these models (e.g. a molecule that kills cancer cells) are then advanced for further evaluation as drug candidates. However, much of this testing is done by screening large chemical libraries in a randomized fashion, with scientists having limited ability to know a priori whether the molecules they decide to evaluate are likely to be efficacious. We therefore wish to develop an AI-enabled platform for molecular search. We wish to allow scientists to select molecules likely to be efficacious in their specific disease model, in a fully automated fashion, to help save time and costs in drug discovery research.

The project consists of developing a solution producing better life expectancy reports than existing third-party providers. Such a solution will improve the overall underwriting process in terms of efficiency, time and accuracy, resulting in better investment decisions.

Vision-based hand tracking is a key component that is raising in importance and will in the future be ubiquitous as part of interaction studies in a broad range of areas such as extended reality (XR), encompassing virtual reality (VR) and augmented reality (AR), and other human-computer interactions (HCI). This project has developed a free and Open Source state-of-the-art vision-based hand pose estimation and tracking pipeline which can be integrated into existing XR frameworks such as Monado (https://monado.dev/).

OpSens develops medical technologies based on a component called microsystem electromechanical (MEMS) whose manufacture is complex. These MEMS are tiny sensors optical pressure gauges placed in guidewires to measure blood pressure at different places in the cardiovascular system, especially in the heart arteries and ventricles. The objective of the project is to allow the early detection of defective MEMS and the optimization of their production through vision and manufacturing intelligence techniques

Development of a technology that will increase the productivity of sewer system inspection work through the creation of new inspection equipment. The camera developed will allow continuous (uninterrupted) inspection of sewer networks. Software will be developed to automatically detect anomalies. This software will perform video analyzes more quickly and reduce the subjectivity of the analysis. The software’s use of the data from this camera will give access to richer information and provide more precise analysis.

Flip Design is a Quebec manufacturing company producing school uniform for over 25 years. The aim of this project is to use artificial intelligence to obtain an accurate sales forecast in order to optimize production and management of our inventory. We have two big periods a year in school uniform sales, however our factory operates annually. Thus, it becomes crucial to foresee our sales to assess the quantity of products to produce by color, size and school.

Creation of an environment based on artificial intelligence models to help decision and risk management on proposals for complex construction projects.

The project is aimed at the design and deployment of automated inspection systems for steel joists for assembly/welding lines. The system must be able to adapt to varying conditions (surrounding lighting, available space, line configuration, movement of joists), know how to recognize the presence or absence of welds and know how to analyze their defects. The use of artificial intelligence and machine vision will serve to reduce or even eliminate the need for visual inspection by a human.

This project will enhance the machine learning capabilities of Sharethrough’s SmartSuite portfolio. SmartSuite is optimizing each step of an impression’s journey through our exchange.

This project uses AI to develop a dynamic purchase and resale price solution, which will be integrated into our application systems for the golf industry. This solution aims to offer the right price in an environment that does not follow the usual market rules.

The Berga.ai project consists of developing tools for predicting supply and demand based on major global trends and known transactional data in the field of recycling.

This project will develop an assisted analysis system, based on state-of-the-art AI/ML modeling, for data from non-destructive testing (NDT) of steel tubes inspected with ECT (eddy current testing) technologies and RFT (remote field test). The system aims to improve the speed and quality of fault detection in heat exchangers, boilers, and other systems using bundles of steel tubes. The solution will benefit asset integrity inspection service providers as well as industrial asset owners. In a broader vision, the project will allow improve the effectiveness of asset management programs and reduce the occurrence of unplanned downtime.

This project aims to develop an integrated conversational engine that will automate a large number of conversations in order to improve the customer experience and internal operational efficiency. The challenge lies in having a language understanding engine specific to LOGISCO with processing and language understanding for Quebec French.

This project consists of optimizing the entry of handwritten and digitized medical billing information. Using artificial intelligence, we want to speed up processing time and reduce errors through automatic document recognition, automatic data transcription, and auto-correction of current data.

The project aims to maximize all the steps necessary for the implementation of an employer branding and SEO 2.0 program. In fact, thanks to the potential of artificial intelligence, the active solicitation of potential candidates through Alithya employees, the intelligent improvement of this involvement and a new employer brand visibility program using NLP techniques will compensate for several limitations of our current program.

To optimize the use of seals on critical equipment for existing and future customers, Robco is developing a Fluid Sealing Optimization Network for Industrial Rotating Equipment, or fSONET (Fluid Sealing Optimization Network), which leverages a secure IIOT architecture, sensors, a detailed performance database and artificial intelligence algorithms to transform traditional maintenance practices of compression packing seals, into scientifically optimized, preventive and prescriptive maintenance.

This project aims to influence the right targets to react more quickly to a suitable rental offer to accelerate the filling speed and to characterize the propensity to purchase of a prospect who initiates a relationship or a request for information for a type of product.

Automation of the powder coating spray line with self-programming robots. The system will be able to find and recognize the items to be painted and will automatically generate robot programs based on their position in space.