Valerii Usov

Product Manager | Business Analyst | IT Consultant

My AI research projects

Hi! I'm Val, a passionate AI enthusiast. My research focuses on app development using AI and natural language, prompt engineering, LLM experimentation, and workflow automation with AI agents.

LinkedIn •

The future of productivity

With AI technology advancing rapidly, keeping up with information has become increasingly challenging for professionals. It's not just about consuming content, but processing it effectively as the volume grows exponentially.

Time management has never been more critical in this digital landscape. The solution might be AI agents built on top of existing apps that become our personal assistants for many routine tasks, saving time and increasing efficiency.

Email management is an obvious and representative example. To spend less time reviewing emails, I built an AI agent that does it for me. Behind the scenes, there's an orchestration agent controlling multiple specialized agents, each trained to perform specific tasks so I don't have to do them myself. For instance, in the attached gif, it overviews emails across my mailboxes and summarizes AI news for the day.

Appreciating the convenience, I believe this might represent our ultimate user interface of the future: simply chatting with an agent, whether through text or voice.
LinkedIn •

Configuring AI Agents: A Business-Centric Approach

AI agents promise to streamline workflows, reduce costs, and open new frontiers of innovation. However, successful implementation requires careful alignment with existing business structures. This article explores a business-centric approach to configuring AI agents—treating them like employees with unique skills, assigning them clear responsibilities, and gradually integrating them into familiar workflows. We examine how to start small, manage risks, and scale over time as agents learn, adapt, and potentially spawn new automated capabilities. While the vision for autonomous agent networks is compelling, it’s important to acknowledge that current technology still requires robust governance, strategic planning, and respect for organizational constraints.
Medium •

Can Small LLMs Working Together Challenge Top-Tier AI Models?

Is it still worth paying for top-tier AI models, or could open-source, locally hosted models now stand as serious contenders? Although metrics suggest the open-source models haven’t quite caught up, what if they could achieve competitive performance when working together as a team?To explore this, I conducted an experiment combining multiple small open-source models to see if their collective intelligence could rival a high-end, paid solution. The results reveal interesting insights into the pot...
Medium •

Developing a Terminator Vision App through Conversational AI

As AI continues to advance, its role in software development is getting more exciting by the day. In this case study, I set out to build an app from scratch that mimics the iconic red-tinted Terminator vision from the movies, using only verbal communication with the AI — and I aimed to do it all within an hour. It was a fun experience with some interesting results that I’m excited to share. Let’s explore the process and the potential it demonstrates.The purpose of this exercise was to explore th...
Medium •

How AI Agents Are Revolutionizing the Way We Work

It’s becoming increasingly evident that AI Agents represent the next significant advancement in artificial intelligence. This transformative technology has the potential to revolutionize our work processes and daily routines. By combining two pivotal capabilities, AI Agents enable models to interact seamlessly with the physical world and operate autonomously, unlocking unprecedented possibilities.AI agents represent a paradigm shift from basic hardcoded or LLM-based automation. By combining deci...