Previous Posts
Goal
- Include more interactive, data-driven, and helpful fishing features in the app.
- Research how to export the app for beta testing.
Workflow
After fixing bugs last week, I realized FishCast was presentable. All of the features in the app were working, and the Terrier data was auto-updating as it should.
I was feeling discouraged about continuing to work with FishCast because I wasn’t sure where to go next in the project’s development. Other interns and staff have encouraged me to make the app more interactive among users. An example would be building a forum where people can post pictures of their fish while capturing the current fishing index conditions. Otherwise, I knew I could focus on tailoring the fishing index to specific fish species to better help users understand the conditions for what they’re fishing for.
One idea that the intern group collectively suggested was for me to get the app in an exportable format. If I could temporarily launch the app to a distinct audience, I could provide them with a questionnaire to gather feedback on how to enhance the app.
After the weekly intern meeting, I first consulted ChatGPT on which fish species to include in the app. The species ChatGPT came up with were largemouth bass, smallmouth bass, walleye, northern pike, bluegill, crappie, perch, trout, and muskie. Then, I asked Claude for suggestions on how to work the different fish species into the app using Terrier data. Eventually, I settled on an overall fishing index and sub-fishing indices for each fish species. This was neatly organized per lake location (a picture of this feature is shown below).
During this process, I asked two anglers I knew to beta test the app. Both candidates were happy to help, with one promoting an idea he had about wind direction and how that affects where the fish are in the lake. Claude came up with an idea to implement a map for each lake location with Terrier’s wind data plotted specifically to the local dimensions of each lake. This is a feature I am still working on to make it more user-friendly.
Overall, this week gave me more goals and a tentative timeline I can use to get as much work done as possible before (and after) we formally present our projects. I am currently planning to send out the questionnaire and the exported projects for some beta testing. Additionally, I will continue developing the new features I added last week.
Prompts Used
When I was discussing app features with ChatGPT, I asked it to provide a list of fish species and conditions that are favorable for catching them. AI gave me a brief rundown of each fish and the conditions, and I pushed what was in the prompt into my project in Claude Code. ChatGPT implemented a sub-fishing index for each fish species, and I have tweaked it accordingly based on my own research. Now, there is a section on FishCast that describes the fishing conditions for each fish species at each lake location.
I worked directly in Claude to incorporate a beta tester’s idea: a local wind-direction forecast for each fishable lake. I worked directly with Claude to add this feature. Claude had an idea to add a wind widget and wind map, emphasizing the shape of the lake and plotting the wind from Terrier on it. So far, Claude has included wind widgets for all the lakes, but the wind maps for each lake have not yet been applied. I am confused about why some lakes have maps and others don’t, but I will work this out with Claude this week.
After I got started on these new features, I have been able to work directly in Claude to make necessary tweaks and run tests to make sure Terrier is running properly within the app. Overall, these newly added features have been easy to work with, and Claude seems to understand how to develop with them.
What Worked
The more I work with AI, the better it gets at understanding FishCast’s purpose and what to incorporate into the app. Adding these features has been very efficient, and noticeably more so than in the beginning stages of FishCast’s development. A whole other side of the app was created last week, and I am amazed at how little I have to tweak these new features.
What Didn’t Work
For some reason, Claude has implemented wind maps for some lakes and not others. I am not sure why this is the case, but I believe this was added before my credits timed out, and Claude couldn’t process the large request in time. This should be an easy fix, and I am hopeful to make this change before the August 14 presentation.
Lessons Learned
Last week, I was discouraged by my progress on FishCast because I struggled to envision the app’s next steps. Thankfully, I have a wonderful team of interns and staff to help guide me. I am grateful for our weekly meetings because they give each of us a chance to discuss what we’re working on, the progress we’ve made, and what we’re stuck on. This past weekly meeting has been incredibly impactful for my overall project goals; I am reassured that the app is useful and coming together as it’s supposed to.
Images/Video
Image 1: Fish species per fishing index scores
Image 2: Specific fishing conditions
Image 3: Wind direction on a lake