Wilfried Rouquet | DeFi Product Designer

Nuant

Merging tools into a cohesive DeFi AI Agent

Nuant build solutions for institutionals and power-DeFi users. They had separates tools around building financial strategies. My goal was to merge them into a seamless experience and create the AI agent layer above it

Problems with the current solution

We have several tools that works on the same object (strategies) but that are separated in terms of workflows. The usage is complex and the adoption of the tools is not optimal

Our goals

The new interface should be integrated and seamless, allowing iterations on strategies, and be easy to share for the internal marketplace we’re building

AI Agent

The engine needs to be manageable by both the user directly or by an agent

Seamless flow

We want to merge our tools suit into a single interface with an intuitive flow to use

Iterations-focused

Users should be able to perform several backtests and simulation and compare different versions

Community-driven

We want to build a community around DeFAI agents

Quick overview of the designed solution

Merged tools

Now backtest and simulation are handled within the strategy object and we have only one menu for everything

Streamlined "Strategy" flow

The whole strategy workflow used to be displayed on several tabs

It’s now displayed in one single page because we want to be able to use the tab-level to manage the backtest and simulation ("Visualize tab")

Backtest and simulation encapsulated within strategy

The most important modification is that now the user can create backtest and simulation in the same interface and within a streamlined workflow

History for iterations

All versions are saved so that the users is able to compare variants and iterate. It's displayed as panel-bar as we use this langage in on others apps

Refine it or publish it to the marketplace

After having performed a backtest, he can either go back to "build" and update its strategy or publish it to the marketplace

Introducing agent

Now it’s time for us to build the layer above with AI agent that will manage the strategy on behalf of the user

Updated marketplace around AI agent

Now our marketplace focus on agent instead of strategies

Agent is now parent of the strategy Vault

An agent is composed of the previous strategy vault, a token, and a personality chatbot

The Agent can create strategy himself

Check this demo video that display the flow

Key takeaway

From this project

AI agent requires two level of analysis

The app need to be usable by both the agent and the user. Also the user need to be able to interact with the agent

Merging tools is risky with limited visual space

The risk when merging tools is to encounter a roadblock where you have too many elements to display or where you'd need a new sub-level that is not possible anymore. So it's especially important to check the long term possibilities

Drop me a message

wilfried@rouquet.io

I look forward to hear about your project !

Substack