Last week, the entire Upstream Tech team participated in a semi-annual hackweek— one of our favorite weeks of the year! What is "hackweek" you might ask? It's when we empty our plates and open our minds to spend five days exploring creative ideas that get pushed to the side in our normal day-to-day work. It gives our team the opportunity to delve into new ideas, work on pet projects, and collab with teammates they might not usually work with. Cool, right?
So what exactly were we hacking? Below you'll find some of the highlights from the past week. Our team worked on everything from incorporating customer feedback into new features to analyzing new markets and testing the usefulness of LLMs. You may not see some of these changes or new features reflected in Lens or HydroForecast right away (or even ever), but all of the projects discussed below contributed positively to our mission and our team's knowledge. Keep reading to learn about some of the results from our latest hackweek.
Since it’s the first page that you’ll experience, having a Lens app homepage that’s helpful and delights is a priority for our team. Building off of feedback, this team redesigned the main page to include helpful information and provide a more streamlined experience. The goal was to provide users with a homepage that displays information that benefits users when they first log in, such as an overview of properties and info about what other users have been up to. The result is a refreshed homepage that still displays all portfolios at the top, but with a new organized tagged system. It also includes new sections such as “Notifications” that displays team activity and alerts from the past week, and “My Properties” for easy navigation. Stay tuned— an iteration of this homepage may be coming to the Lens app soon!
The ability to share what you’re seeing when monitoring is a core feature of Lens. Through reports, screenshots, GIFs and more, we know that Lens users appreciate the ability to share what they’re working on. This hackweek team took things to the next level. The goal was to create new functionality that allows folks to share interactive examples of what they’re seeing in Lens. The result is a new way to show and share interactive Lens examples through either a live link with some available functions, such as Compare Mode, or an embedded map that can be added into public web pages. We’re super pumped about this new ability which will expand storytelling capabilities for our team and all Lens users!
Having a smooth set-up is an important part of onboarding HydroForecast customers. This team decided to focus on shifting the delineation process from a backend experience to a frontend one. The goal: self-service basin delineation, in which all that’s required is to drop a point, submit it for delineation, and then watch as your product page is updated. As the team iterates on this hackweek project, we expect this to open the door to quickly and easily adding other HydroForecasts products, such as virtual gauges and Long-term, to existing customer accounts.
There are so many unique use cases and questions that can be solved with Lens. That’s where our Lens in Action series comes in: it’s a collection of short videos that show unique examples and interesting tips and tricks. During Hackweek, the team put together 10 new videos that show off interesting examples such as using Analyze Area to learn about snowpack, monitoring for invasive plant species, and identifying change using Planet Basemaps and Compare Mode. You can view all Lens in Action examples here.
We have a lot of operational forecasts to review daily, and we want to ensure we’re alerting users about interesting events when they occur. This team recognized several problems: it’s time consuming to review all forecasts manually and it’s not a scalable solution when customers are interested in different types of events. As an experiment, the team tested giving data to a Large Language Model (like ChatGPT) to see what it did with our forecasts. There were several interesting findings: GPT-3 had a tendency to say illogical things, even when given clear prompts and data, but GPT-4 gave a coherent and logical response when given the exact same prompt and data. The team found that there was still a discrepancy between what GPT-4 classified as an “interesting event” vs. what the team would say, but this could be improved with refined prompts. There is still more to learn, but the team walked away with interesting insight about LLMs and some ideas for the future.
This team set out with the goal to provide customers who are looking to get an even deeper understanding of the data available in Lens with the ability to customize the visualization of some data. The team prototyped some work that allows users to dynamically change the data we’re rendering on the fly. The ability to manipulate the tiles from the frontend will give Lens users a deeper understanding of the data they’re interested in and will allow them to use Lens for more advanced geospatial use cases. We're excited to iterate on this feature and potentially bring it to Lens soon!
Building on feedback and what we know about how non-profit organizations use parcel data, this hackweek project produced a new type of report in Lens that’s centered on easily providing parcel owner data in one report. From mailing fundraisers to running outreach programs, learning about the parcel owners alongside the conserved properties they manage is helpful for these organizations. The new Parcel Owner Report includes a clear list of parcel owners along the edges of their conserved property, and is generated in one-click instead of requiring users to click around their property to find the names of parcel owners one by one. More iteration to come, but we expect this report to be a useful tool for non-profits!
It currently takes several hours minimum to get weather data at any new site, so this team asked the question: what if it was instant? The goals of this hackweek project were to increase the speed in which we get data, move towards a fully automated customer onboarding process, and make a public weather API product. Not a small task! The team created a new archive that was organized for speedy access and built several features on top of it, such as a way to input longitude and latitude and then receive the latest weather forecast (temperature, precipitation) in the form of a gif. This has multiple potential uses, including the ability to provide open source data to the wider community, and we’re excited to work on a future state!
This team started with a couple of core questions: How might Lens support environmental justice outcomes, and how can we actively consider this in our product decision process? One of the most exciting outcomes of this team’s work over the course of the week was the addition of new data from Native Land Digital, an indigenous-led Canadian non-for-profit. Integrating this dataset into a new overlay in Lens will simplify customers’ existing community outreach workflows and offer new ways to connect. Our team is planning to meet with Native Land Digital to ensure our use of their dataset aligns with their mission before we launch the new overlay. Going forward, we will ensure that our team and our customers continue to take a socially responsible approach to conservation.
We recognize that GIS and remote sensing methods are taught and utilized at universities around the world, and we believe there is an opportunity for individual researchers to save time and quickly access the data they need with Lens. This team put together new documentation and use case examples that demonstrate how Lens is useful to this audience, and we’re looking forward to welcoming academic researchers to the diverse community of Lens users!
---
And that’s not all! Other team members worked on important internal tasks, such as updating process docs and creating new Slack bots, that will make things smoother and more efficient for the whole Upstream team. All in all, we had a fantastic week full of innovation and creative ideas. Shout out to the whole team on the incredible work that came out of this week! We’re already looking forward to the next one 😎
