Progress is being made every day. Generative models produce text, images, and music with results that would have seemed impossible not so long ago, and some studies suggest that the emergence of new object recognition methods could speed up computer vision application development. Artificial intelligence (AI) was conceived many years ago, but it reached the general public only in 2022, thanks to the rise of ChatGPT or Copilot.
Models can be asked to classify (for example, select documents by type), summarize (summarize extensive texts by extracting the key ideas), reason (problem-solving), plan (follow a series of steps to reach a goal), and, of course, generate content.
Therefore, content generators are the most widespread solution. However, self-sufficient agents appear, i.e., bots that are given an objective and are capable of working towards it, defining the tools they will need and incorporating others as they execute it as the situation requires. These chatbots identify the people who interact with them to generate a hyper-personalized experience.
The era of continuous discovery
What is our challenge at Making Sense? To evaluate how these innovations allow us to work better internally and, at the same time, to be constantly on the lookout to see how we can use them to create value for our customers.
For us, discovery is no longer a phase but a habit. We listen to ideas, think about challenges, improve processes, add features, or design new products. We rely on two pillars: quantitative and qualitative information and applied data analytics.
For example, one of our clients, an agricultural software solutions provider, has a tool that is a command line that the user can use to execute actions, in particular, to run specific queries or visualize various reports. It is a valuable resource since about 10% of users use it daily to generate reports.
However, not everything is so easy. To run the queries, it is necessary to understand syntax, which is also complex. Analytics showed us that it is the option most chosen by users when obtaining information from the system and, moreover, that 20% of the executions of this command line failed, mainly due to syntax-related issues.
Expectations come true
Generative AI helped us solve the problem. Today, users can run their queries in natural language (and in any language), also allowing us to significantly expand our reach and enabling a more significant number of users to take advantage of the command line for their daily tasks. People can access more accurate and faster information, without errors, to make better decisions.
This is just one of the many applications in which we are already exploring and exploiting the power of AI. We have also used AI to detect the best time to guide investment strategies based on a comprehensive parameter analysis, or to predict the quality of future sprints in a development process, to name a few examples. The possibilities are limitless.
In conclusion, it is not just about implementing an AI model or identifying a specific use case for a specific problem but keeping our minds open and our teams in a continuous search mode. The opportunity to extract value from this technology may appear where it is least expected.