Meet Mura, an LLM built for blockchains.
ROLE
Product design
COMPANY
OpenBlock
YEAR
2024-2025
Led the design of Mura, creating an intuitive interface to transform on-chain data into actionable insights through real-time reasoning.
NORTHSTAR
Mura delivers a seamless blockchain knowledge platform, turning raw data into actionable insights through real-time simulations and simplifying complex on-chain activity for informed decision-making.
PROBLEM FRAMING
OpenBlock data analysts frequently address one-off inquiries from blockchain protocols regarding the health of incentive programs, which consumes significant time and manual effort. The primary challenge lies in productizing these types of questions to efficiently generate actionable insight.
MINIMUM VIABLE PRODUCT (MVP)
The fastest way to test Mura as an MVP was through a Slackbot, allowing us to quickly experiment with the model’s ability to query blockchain data and generate meaningful insights.
REDUCED FRICTION
I reduced onboarding friction by simplifying the language from “Connect Wallet” to “Sign In” and adding support for Google and Twitter logins. This made the experience more approachable and intuitive, resulting in a significant increase in sign-ups.
USER RESEARCH
I conducted user research to understand the needs of our primary users, data analysts. I tested a chain selector tool that lets users multi-select chains. Still, customers wanted deeper visibility—focusing not just on which chains were available, but also on the specific data within each chain.
USER RESEARCH TAKE AWAYS
Through user research, I learned from customers’ questions, such as, “If I need a specific metric, can I tell if the LLM has it before querying?” This insight led me to reimagine how data analysts interact with tables, transforming them into an intuitive data explorer.
Customers can drill down into detailed data, interacting with tables through Mura to refine insights and make complex blockchain data more accessible for analysis.
REASONING EXPLORATIONS
During early testing, certain data-intensive queries took over two minutes to process. To improve the experience, I explored different designs for reasoning states after a question had been asked. I ultimately adopted a chain-of-thought approach, enabling Mura to break down queries step by step, enhancing transparency and reassuring users that the model is actively working.
I also split the models into a quick search for faster results and an on-chain reasoning mode, which took longer but provided more accurate blockchain data.
REVIEWING DATA
Customers could review the SQL code from their submitted queries and enable a split-panel view to switch between the code and dynamic data visualizations. This enabled them to review the data, publish it, and easily share insights with others.
BRANDING
I took a holistic approach to branding, spanning logo design, marketing, and seamlessly weaving the brand identity into the product experience and marketing assets.
IMPACT & LEARNINGS
KEY RESULTS
Streamlined the onboarding process, making it more intuitive which resulted in higher sign-up rates.
Streamlined the onboarding process, making it more intuitive which resulted in higher sign-up rates.
Reduced internal workload by enabling OpenBlock analysts to spend less time answering protocol questions, allowing them to focus on higher-value analysis.
+76%
Increase in sign-ups.
-20%
Reduced internal workload
