Hi AgroSmart, thanks again for joining the DDAccelerator! To kick things off, could you start by introducing AgroSmart – what exactly does your product or service do?
AgroSmart is an innovative system developed with the goal of driving digital transformation and enhancing agricultural production in Bosnia and Herzegovina through the application of Artificial Intelligence (AI). Our solution enables farmers to quickly, easily and accurately estimate wheat yield. After capturing a photo of a field sample, the AgroSmart system automatically processes the image and calculates the number of wheat ears, which is a key indicator of yield. Based on this, the user receives a reliable estimate of the total yield.
Our aim is to make this technology accessible to all farmers, without the need for expensive equipment or advanced technical knowledge. The system is designed to be intuitive and efficient, offering users the choice between submitting their own data or using our professional image capture service. In doing so, AgroSmart not only simplifies daily agricultural processes but also contributes to the sustainable use of resources, cost reduction and increased productivity, making it a valuable tool for anyone seeking to modernize their farming practices.

What inspired you to develop AgroSmart, and how does your AI technology work to improve the accuracy of wheat yield predictions?
‍The development of the AgroSmart system was inspired by the real challenges faced by farmers in Bosnia and Herzegovina, particularly the lack of access to modern technology, precise data and decision-making support. Traditional methods for estimating crop yield are often inaccurate, slow and require expensive equipment or expert knowledge, which are not always available. Our vision was to create a solution that is simple, accessible, and reliable, a tool that any farmer can use, regardless of the size of their farm or their technical expertise.
AgroSmart relies on Artificial Intelligence models trained on hundreds of labeled images of wheat fields collected directly from the field. Our AI algorithms are designed to recognize wheat ears based on photos submitted by users via mobile devices, cameras or drones. Once the image is analyzed, the system automatically calculates the density of wheat ears per square meter, indicator for estimating total yield. This process allows for fast, accurate and locally relevant predictions that help farmers make timely decisions, such as applying additional nutrients or adjusting agronomic practices with the goal of increasing productivity and reducing losses.

How do farmers engage with the platform, and what is the user experience like from their point of view?
Farmers can engage with the AgroSmart platform in two simple ways, tailored to their needs and available resources. The first option allows them to participate independently, by using a smartphone or camera, they take photos of selected wheat field samples according to clearly defined guidelines and upload the images to the system. AgroSmart then automatically analyzes the images and provides an estimated yield. This approach is designed to be intuitive and easy to use, even for those without prior technical experience.
The second option involves support from the Verlab Institute’s expert team, who visit the field, use professional equipment (digital cameras or drones) to capture images, and then process the data through the AgroSmart system to deliver results to the farmers. From the user’s perspective, the experience is fast, practical, and requires no complicated steps.

Bringing artificial intelligence into agriculture holds great promise but also comes with significant complexities – what technical, practical, or even cultural challenges have you encountered in developing and implementing your solution, and how have you approached overcoming them?
Introducing artificial intelligence into agriculture offers tremendous potential, but it also comes with a range of challenges: technical, practical, and cultural. One of the initial technical challenges was ensuring access to high-quality and diverse data for training the AI models. Collecting such data in the field requires careful organization, standardized recording conditions, and close collaboration with farmers. Additionally, the terrain itself posed difficulties due to inaccessible plots, steep slopes, and varying soil types, which made data collection and image quality more complex. To overcome these challenges, we developed clear data acquisition protocols, tested multiple methods (using cameras and drones), and worked closely with local farmers to ensure that the data we gathered was realistic and relevant.
From a cultural perspective, we encountered a certain degree of skepticism toward new technologies, especially those that fall outside traditional agricultural knowledge and practices. To address this, we organized educational workshops, involved farmers directly in the testing process, and demonstrated tangible benefits on their own fields. Over time, as they saw the system's accuracy and usefulness, resistance gradually diminished.

We’re curious to hear whether agriculture and innovation in this field have a personal connection for you or your team – does this topic reflect your background or previous experience?
The AgroSmart project brought together a multidisciplinary team of experts from both natural and technical sciences, including engineers with backgrounds in biomedical engineering, data science, and artificial intelligence. While this was our first project specifically focused on the agricultural sector, our team has extensive experience in applying AI technologies across various domains such as healthcare, environmental monitoring, and industrial systems.
What connected us to agriculture in this case was the clear opportunity to apply our expertise to a sector that is vital for sustainable development but often underrepresented in the field of digital innovation. We recognized that the agricultural landscape in Bosnia and Herzegovina holds great potential, yet still faces challenges that can be effectively addressed through smart technologies.
As the National Competence Center for High-Performance Computing (HPC) in Bosnia and Herzegovina, we also possess the infrastructure and technical capacity needed to develop and train advanced AI models.
AgroSmart allowed us to bridge that gap-applying advanced AI methods in a practical, field-based context, while closely collaborating with farmers and agricultural stakeholders. For us, this project was not just about developing a tool, but about proving that artificial intelligence can be tailored to meet real-world needs in sectors traditionally overlooked by high-tech solutions.‍

Reaching traditional farmers who aren't yet familiar with digital or AI-driven tools can be challenging. What’s your approach to building their trust and helping them adopt new technologies?
Reaching traditional farmers who may have limited exposure to digital or AI-driven tools requires a thoughtful and human-centered approach. Our strategy has focused on building trust through direct engagement, education, and clear demonstration of value.
First, we conducted field visits and one-on-one sessions with farmers, where we listened to their needs, concerns, and current practices. Rather than presenting AgroSmart as a high-tech solution, we framed it as a practical tool that could simplify their work and help them make better decisions with minimal effort.
We also organized educational workshops tailored to non-technical users, showing them exactly how the system works, step by step. By involving them in the data collection process and letting them see firsthand how images from their own fields could translate into reliable yield predictions, we turned skepticism into curiosity and eventually, acceptance.
Crucially, we avoided overwhelming users with technical jargon. Instead, we focused on outcomes they care about: saving time, improving yields, and reducing costs. This result-oriented communication helped them see the immediate benefits of using AI in their daily work.
Trust was built not only through outreach but also through consistency, reliability, and a solution that respects their time, knowledge, and way of working.

As you continue refining your technology and working with farmers, have you started exploring how your solution could be adapted to other crops or scaled to different agricultural regions?
While AgroSmart was initially developed with a specific focus on wheat, from the very beginning we envisioned the system as a scalable platform adaptable to various crops and regions. The core of our solution lies in the use of AI-based image analysis, which can be applied to other types of crops such as corn, barley, or even fruits and vegetables.
We have already begun preliminary research into adapting our models for different plant structures and growth patterns. This includes collecting and annotating new image datasets and exploring how agronomic variables change across crop types. We’re also in discussions with agricultural experts to ensure any future adaptations are scientifically grounded and practically useful.
In terms of geographic scalability, we recognize that soil composition, climate, and farming practices vary significantly between regions. That’s why we’re designing AgroSmart to be modular and adaptable, allowing for regional calibration of models to ensure accuracy in different agricultural contexts.
Ultimately, our goal is to build a robust AI-based platform that can support farmers beyond wheat and beyond Bosnia and Herzegovina, empowering data-driven agriculture across multiple crops, regions, and scales.

Do you aim to integrate AgroSmart with existing farm management systems or other agricultural technologies?
Yes, integration with existing farm management systems and complementary agricultural technologies is a part of our long-term vision for AgroSmart. We recognize that farmers increasingly rely on a variety of digital tools, whether for planning, irrigation management, crop protection, or logistics and we want AgroSmart to seamlessly fit into that ecosystem rather than function as a standalone solution.
From a technical perspective, we are working toward developing interoperable APIs that would allow AgroSmart to exchange data with popular farm management platforms. This would enable farmers to import geolocation or historical crop data into AgroSmart, and conversely, export yield predictions and analysis back into their broader farm management system for strategic planning.
We also see strong potential for integration with sensor networks, satellite data, and weather monitoring systems to enhance the accuracy of our AI models and provide even more context-aware recommendations.

At this stage of your journey, which types of partnerships or forms of support would help you move forward most effectively?
The most valuable support would come from investors who recognize the potential of AgroSmart and are ready to help us scale its impact, as well as from farmers who are open to adopting this technology in their daily practice. Their involvement is key to ensuring broader use, validating the benefits on the ground, and creating a ripple effect that drives real change in agriculture, both locally and beyond.

What were your main expectations going into the DDAccelerator – and have they been met so far?
Our main expectations were to gain practical insights into scaling digital solutions, build meaningful partnerships, and refine our business model through expert feedback. So far, these expectations have been met, because the program has provided valuable tools, guidance, and networking opportunities that have helped strengthen our AgroSmart initiative. We’ve been able to validate key aspects of our approach, improve how we communicate our impact, and connect with a supportive community of innovators.‍

Finally, how do you see AgroSmart’s impact in the future, both in your region and beyond?
In the future, AgroSmart has the potential to become a key driver of sustainable agricultural transformation, not only in our region but also in other countries facing similar challenges. By democratizing access to AI-powered tools for yield prediction and crop management, AgroSmart empowers small and medium-sized farmers to make data-driven decisions, optimize resources, and increase productivity. In Bosnia and Herzegovina, it can significantly contribute to food security, rural development, and climate resilience. Beyond our borders, AgroSmart can serve as a replicable model for digital innovation in agriculture, particularly in developing regions where affordability, simplicity, and scalability are critical.

We're truly glad to have you as part of the DDAccelerator. Wishing you all the best as you continue to drive innovation in agriculture – we're excited to follow your journey and future successes!