Designing AI Agent tools at Maven AGI
Revamping the Agent Designer experience by crafting user-first designs that balance technical complexity with intuitive interfaces, helping users engage with and optimize AI tools with ease and confidence
Context overview
During my time at Maven AGI, I worked on two main areas: improving the existing agent products (like the Copilot agent) by incorporating user feedback and enhancing core product functionalities, and leading a comprehensive overhaul of the Agent Designer — the main tooling area where users train, tune, and test their AI agents.
The challenge was to create a more intuitive, streamlined experience that reduced friction and made the product more self-serve, without heavy reliance on customer support teams.
My role and approach
I was tasked with performing a usability audit of the existing system, understanding key pain points, and proposing a design and experience overhaul to make the product experience more intuitive, intelligent and user-friendly.
TO DO THIS, I:
Conducted stakeholder interviews to better understand the user journey and gather direct feedback from customers.
Led a deep dive with engineers and leadership to align on the product’s vision and technical constraints.
Focused on redesigning the system from a user experience perspective, aiming to enhance usability, reduce cognitive load, and increase transparency in the AI-first experience.
Impact
The rehaul of Agent Designer had a significant positive impact on the product's usability:
Reduced cognitive load: By streamlining the UI and removing redundant features, users could more easily navigate the platform, leading to faster setup times and fewer errors.
Increased user autonomy: Features like Ask Maven and the pre-conditions plugin empowered users to troubleshoot and manage their agents independently, reducing the need for customer support intervention.
AI-first experience: By integrating AI-powered insights, such as the recommendations in the test suite and AI-driven conversation feedback, we were able to enhance the user's understanding of the AI's behavior and decisions.
Improved user satisfaction: We received positive feedback from customers who reported a better understanding of how the bot works and more confidence in using the tool. Customers were also able to make data-driven improvements to their agents without the need for constant hand-holding.
Outcome
The final design not only simplified the product but also aligned with Maven AGI’s philosophy of augmenting human performance with the right amount of AI integration, giving users the right tools to take control of their AI agents while maintaining a transparent understanding of bot reasoning and human-centered approach to designing features.
This project reinforced the importance of aligning with users and stakeholders early in the process, especially in a startup environment with rapidly changing priorities. By focusing on first principles design and leveraging user feedback, I was able to create a product that felt intuitive and efficient while also maintaining technical robustness.