An interface for Decentralized Market Making with Hummingbot

Charlie Ellington
Deep Work Studio
Published in
4 min readApr 26, 2019


The Hummingbot team asked us to explore how a market making interface could help solve the liquidity problem in decentralized finance. We jumped at the chance to help solve a key problem for Open Finance and DeFi.

A screenshot from the prototype produced in less than a week

The results are nothing but remarkable with the sheer quantity of work and insights provided second to none.

In a week, the Hummingbot team were able to:

☑️ Validate the product and features before a costly building phase.

👩‍💻 Get real user feedback on the solution.

🤩 Rapidly learn if they were building the right thing.

The process was a GV Design Sprint. If you’re unfamiliar with the process we wrote a blog post highlighting why Design Sprints are great for blockchain teams (TL:DR often blockchain teams don’t know if they’re building something users want until after a costly and time consuming development stage).

Here’s the end prototype result after a week:

Hummingbot Interface Prototype

How we did it…

Defining design challenges and generated solutions

We worked together with the Hummingbot team to both gain a better understanding of their goals and problems.

By defining the long term goal in specific collaborative steps. We were able to set the Sprint Questions. The process avoids wasting valuable time on miscommunication or lengthy traditional agency or designer research stages. We got straight to work.

From this we created a full user journey. Looking back at the goals we could easily find the area of the greatest business value.

We had the key decision makers working synchronously and asynchronously

The Sprint team consisted of Mike the CEO of Hummingbot (CoinAlpha), Yvonne Zheng, Carlo Las Marias, Ryan Chao and Martin Kou. We ran the sprint remotely using a format developed over the years by Andrej.

On the Deep Work Team we had Andrej — facilitating the Sprint having run over one hundred with teams such as Lego, N26 and Axa. And Jim — A UX designer with specific skills taking a command line tool to a graphical interface.

By having the entire team and a focused skillset we made sure we were maximising the results. The teams were aligned and compressed the normal back and forth between clients and designers into a short space of time.

Sketching Competing Solutions

The entire team — client and us — sketched their solutions to the Sprint questions. By moving beyond the usual decision makers, we got a full range of ideas.

Probably the most intense session — the storyboard outlines the exact story we’ll test with users. We took all the work and votes so far to create the stages.


The goal of every sprint is to rapidly build a high fidelity prototype. This is not a “wireframe” or a “paper prototype”, it looks and feels like a real product. We always make sure that the test simulates the actual product because then and only then you are getting the authentic and emotional feedback needed to learn from the test.

With Hummingbot — Charlie joined Andrej to rapidly create the following designs:

You can try out the clickable prototype here.

Designed with a mixture of Sketch and Figma.

User Experience Interviews

By gathering the feedback of users we proved or disproved the hypotheses and answered our sprint questions. By testing with multiple users we got real qualitative data with clear trends.

With Hummingbot we gained clear insights to how we could improve the overall product strategy. We found where we should focus our design efforts based on the user needs.

The Results

After testing with real people, we quickly found that the proposed solution was the way forward for Hummingbot.

We know where the user value is and what features or ideas should be improved, discarded or emphasized in the next step of the design and development cycle.

All this was achieved in a week before a single line of code was written. This has saved the Hummingbot team huge amounts of time, capacity and money. Compared with if they had hired a designer for a lengthy UX research period or built the tool themselves before gaining feedback.

By being focused we’ve produced a huge amount of data, results and designs.

Design Sprints check that you’re building the right thing with real user feedback in days not months!



Charlie Ellington
Deep Work Studio

Product Design and Business Strategy. Helping build blockchain products for the next generation of users. 😀🎨⛓️🌊🏄