Research

Currently, I am cooking AI related concepts in Dev / Designer Tools, New UIs, Multi-Agent Systems… more to come soon.

We spent the past 30 years building lots of software. But in the process the UI paradigms have become very bloated (think Microsoft Word menu buttons and many 3D modeling apps), with a majority of the functions not necessary for the user at any given time.
Rise of AI has potential to change all of that. But the prompt alone will not be the answer.
The next 10 years will be a great time for AI-first teams to rethink UI. My guess is most of the software and website UIs will be completely re-done and in a more adaptive and contextual way.

The concept of AI agents have been around for as long as the AI field was conceived decades ago. However, making real agents work was elusive, until now. I think we have lacked models that could “reason” and act on our behalf, so most MAS systems in the past were mainly simulations.
New projects are pop’ing up as we speak where a team of agents collaborate to do complex tasks, from stock trading, content creation, etc.
The agentic frameworks we have today (LangChain / LangGraph / Autogen / CrewAI) will continue to evolve, as they are still early and not yet “easy.”
I am most interested in systems that require agents to negotiate, for example, different agents all acting for their respective owners to compete for a digital ad slot or for GPU compute slots.