From Months to Minutes: The Evolution of Computer Vision

Computer vision has promised to transform how organisations understand and respond to the physical world. By enabling software to interpret images and video, it has created new possibilities across manufacturing, logistics, construction, retail, healthcare and many other industries.

Yet for much of its history, computer vision has remained difficult to build and deploy.

Early computer vision systems relied heavily on manually defined rules. Developers had to tell the software exactly what visual features to look for, making applications expensive to create and difficult to adapt. The emergence of machine learning and deep learning changed what was possible. Systems could detect objects, recognise patterns and analyse increasingly complex scenes with far greater accuracy.

However, the development process remained highly technical. Creating a computer vision application typically required data collection, annotation, model selection, training, testing and custom integration. Even relatively straightforward ideas could take months to move from concept to a working application. Organisations needed specialist machine learning expertise, significant budgets and access to large quantities of relevant data.

As a result, many valuable ideas never moved beyond the planning or proof-of-concept stage.

The next evolution of computer vision is changing that. Advances in multimodal AI and visual reasoning are making it possible for systems to understand context, relationships and activity within a scene, rather than simply identifying predefined objects.

This shift has led to the launch of Viso Now.

Viso Now is an agentic computer vision platform that allows users to build working AI vision applications using natural language. Instead of collecting datasets, training models or writing complex code, users describe what they want the application to understand.

They can upload an image or video, enter a prompt and receive a working application in minutes.

A safety team might ask Viso Now to identify when a pedestrian moves too close to a forklift. An operations team could analyse footage to uncover bottlenecks or periods of inactivity. A quality team might build an application to spot defects, contamination or missed process steps.

The median time to create a first application is just three minutes.

This represents more than a faster way to build computer vision. It changes who can participate. Engineers and AI teams can test and deliver ideas more quickly, while operations, safety, quality and innovation teams can explore applications without needing to begin with a lengthy technical project.

Viso Now also gives organisations a practical route from experimentation to real-world deployment. Teams can start in the browser, validate an idea with existing footage and collaborate through a shared workspace. When they are ready, successful applications can be scaled through Viso Suite across live cameras, edge devices and multiple locations.

Computer vision is moving from a specialist technology to an accessible tool for solving everyday business challenges. The launch of Viso Now marks an important step in that evolution, turning ideas that once required months of development into working applications created in minutes.

The opportunity is no longer limited by how quickly an organisation can assemble a machine learning project. It begins with a much simpler question: what would you like your cameras to understand?