Refuge.

Refuge.

Refuge.

Work-in-progress

Wildlife non-profit management solution brought to life by ai

Refuge is a personal project based on my volunteer work with the Austin Bat Refuge. It's goal is to support the non-profits mission by increasing organization, efficiency, data quality, and resource management capacity. While also creating replicable systems.

Project scope

  • Design and build comprehensive organization management solution

  • Validate and test effectiveness

  • Long term - volunteer mobile app

Tools

  • Interview scripts, notes, Figma

  • V0, Material, Untitled UI,

  • Cursor, VS Code, HTML, CSS, JS, React, Tail-wind, Git, supabase

Methods

  • User interviews, field studies, collaboration w/ experts

  • Wire-framing, design systems, hi-fi Figma mock-ups, ai prototyping

  • Outside engineering reviews

My role

  • Research

  • Design

  • Development

Duration

  • 2 months

  • Work-in-progress

Date

  • Began 01/2025

Tools

  • Pen & paper, Figma, Figjam, Lucid

  • Excel, Amplify

  • Jira, Slack, Aha!

Duration

  • ~1 year development: technical implementation & testing based refinement

Date

  • Dec 2022 - Feb 2023, 10 weeks initial research and design

A chance encounter

In 2019 my partner and I met a group of bat researchers in the wilderness while back country camping in Big Bend National Park. They invited us out that night to catch and release bats with them in support of a study they were doing. They were led by Merlin Tuttle, a famous bat researcher widely regarded as the leading expert in the field.

In 1986 Merlin was able to leverage a scientific study in advocacy of Austin's Congress Avenue bat colony proving their positive environmental and financial impact on the region. This earned the colony a protected status locally.

Back at home

Our encounter in the Chihuahuan Desert left us wanting more. Really my partner's newfound awareness of the bat research community quickly led her to a local non-profit, the Austin Bat Refuge.

Austin Bat Refuge is a 501 c non-profit, with permitting through the United States Department of Agriculture.

Hilary has been volunteering with the Austin Bat Refuge for nearly 4 years at the time of writing and currently servers as President of the Board. My involvement largely stems from her continuing commitment to the project.

"Our team consists of veteran bat rehabilitators and educators with extensive experience in urban wildlife issues."

Framing an opportunity

Austin Bat Refuge's goal is simple, rehabilitate injured bats and educate the local community about bats. In practice these people work non-stop, it is a true labor of love taking in and caring for injured animals, constant community outreach via events and social media, managing volunteers, and endless fundraising and political advocacy. The founder is even training an LLM with the local bat flight data he records.

Over the course of several years I have gotten to know the team and their work first hand. I see them use their tech stack in the field in real time and have regular conversations about their needs and challenges.

What if a small non-profit could afford a full-scale custom management solution?

What if a small non-profit could afford a full-scale custom management solution?

What would a flexible animal rehab management solution look like?

What would a flexible animal rehab management solution look like?

Intial concept mock-ups

My thought process

I began with the dashboard, which is not something I'd normally do for this type of application, but I'm more familiar with this problem space than normal. The dashboard is based on field research, years of observation, and interviews with core team members. It captures the team's daily workflows.

The dashboard is meant to serve as a vision for the product. Later iterations will be informed by the modules as they are fleshed out.

The dashboard has been strategically paired with a data table. The data table is simple and practical. It serves an a pragmatic entry point to developing the product. The data points and their information architecture area based on interviews with the core team. This will be the beginning focus of my ai based development work. I also may try to code this myself after working with Ai.

Let's just try Ai…

My Ai development work is experimental and diverges somewhat from the UX heavy work around the Austin Bat Refuge. It focuses on understanding the limitations and possibilities of ai driven development and where it fits in the over all software development process.

If a final product is achievable, that's the goal. If a testable prototype is the limit that's not a failure and it has a place. If it really is just 15yr old CRUD technology I'm here to learn that for myself and share my experience. While I'm by no means a hype-man, I believe in experimentation and first-hand experience.

At the time of starting this I have a basic understanding of HTML, CSS, javaScript, React.js, and Git.

Case study title may change as the project progresses.

1st V0 build with Supabase integration

Work-in-progress

Notes

  • Really mixed results so far. Both impressive and unstable.

  • Component based approach seems unrealistic in V0

  • Supabase integration was seamless and effective.

  • Prompting is expensive and tedious, really looking to dial in efficiency.

  • 20 prompts total shown here, results are not a functional prototype

  • More to come

2nd V0 attempt

2nd V0 attempt

Notes

  • 20 prompts again

  • This build threw errors related to the supabase integration every go

  • This build functions locally. The new bat flow works.

  • This has mid-fi prototype potential in some contexts, but the lack of visual accuracy makes it less useful in others

  • I am asking for significantly higher visual accuracy than I'm getting

  • I'd like the build to be a little cleaner before asking an engineer to review. I'd like to have a minimum level of security checks before attempting testing with users among other things.