Sift
Developing an open-source platform for resource sharing and collaboration amongst Forest Restoration Practitioners in partnership with Collaborative Earth.
About the project
A team of four designers from California College of the Arts, collaborated with Collaborative Earth on the capstone project. Over the course of 8 months, we conducted extensive user research and developed a tool for practitioners to gather large datasets.
Year
2024 | 8 Months
Tool & Skills
Figma | UX Research, User Testing, Visual Design, Design Systems
Team
Kirtana Kannan, Priyam Shah, Rishma Bora, Anusha Thalnerkar
Mentors
Rochelle Ardhesher, Marc O’ Brien

What does Collaborative Earth do?

Consisting of 2,400 members, including volunteers, scientists, engineers, and other team members, Collaborative Earth operate in various ecological regeneration labs to co-create pathways for social and ecological regeneration.
One of the labs specializes in 'Assisted Forest Regeneration'

Challenge

Sifting through sparse information is a challenge for forest regeneration experts

We received a brief from Collaborative Earth with the mission to review and synthesize data due to the abundance of unpublished research, documents, and case studies, some of which are in non-English formats.

outcome

AI-integrated open-source platform for forest restorations, enabling quicker review of research papers.

The project resulted in an open-source repository that uses artificial intelligence to enhance planning and collaboration by providing insights, community knowledge, resources, and mapping tools.

overview of the solution

Smart Search Engine

Searching for case studies and publications based on the location in the implementation stage provides valuable and accessible resources.

AI-Powered Resource Review

Users can upload resources to quickly generate and analyze insights by leveraging AI, enabling more efficient paper reviews.

Annotate & Collaborate

Practitioners can save resources in the workspace, collaborate with team members, and have discussions all in one place.

Interview insights

What did the Practitioners have to say?

“For restoration to be successful, it should improve people’s livelihoods - primarily financial.”
Restoration Ecologist at Collaborative Earth
“I would note that use of a mapping program has been a bit of an obstacle maybe in certain projects.”
Organization Lead at Collaborative Earth
“Building a centralized digital tool will be really helpful for the people to access”
Restoration Ecologist at Collaborative Earth
how might we
Provide standardized information and facilitate collaboration among restoration practitioners?
behind the scenes | Exploring ideas

Forest restoration is a specialized field that demands a thorough understanding of its complexities.

The term 'Practitioner' include a variety of roles, such as Restoration Ecologists, Restoration Practitioners, and Landscape Designers, many of whom fall under this broader category.

What didn't work out as expected?

Using the insights, we brainstormed ideas centered on information accuracy and progress tracking for volunteers and practitioners, generating four distinct directions.

Validation and post testing

Exploration #1

The search bar was designed for all-in-one functionality, but stakeholders preferred a feature-rich, transparent version with essential functions and AI integration for enhanced search refinement.

Exploration #2

The tool offers links, resources, and case studies as a repository. They cannot host their projects due to insufficient resources and the need for additional setup from the organization.

Exploration #3

AI will be integrated into the platform's backend to generate content, summarize papers, and account for users' locations, but the smart search engine will not feature conversational design which is better suited like ChatGPT.

Exploration #4

Volunteers are unable to upload projects because they lack the necessary training and expertise for paper publication.

Introducing SIFT.
One tool is all you need.

Multi-search engine

We implemented a multi-purpose search function, enabling users to both search and quickly extract insights by uploading documents, research papers, or policy documents.

AI-powered search is an advanced search bar helps them to find relevant and well-cited sources.

Adaptive filtering

Recognizing the distinct challenges of restoration in each location, Sift features a filter panel that allows practitioners to easily find the most accurate information and refine their search.

Collaboration

The collections enable collaboration among team members and organizations, where they can track the types of resources saved, the number of files stored, and the count of participating members.

Customize tags

Practitioners can create their personal space, using customized tags as organizers for clear tagging and labeling when saving resources.

Smart Insights

Smart insights are generated through LLMs analyzing thousands of vetted resources. The tool produces concise summaries, making it easier for practitioners to review the material quickly.

AI-integrated insights

Using the LLM model, this tool offers dual functionality for searching and obtaining quick insights.

outcome

Easily search documents and receive smart summaries

Searching for case studies, papers, and publications by location or implementation stage can be a valuable and easily accessible resource.

Gain instant insights by uploading a file, enabling you to quickly extract valuable information.

AI-powered insights help practitioners use search engines to find relevant facts and figures.

Based on the uploaded document, they can access a variety of content, including case studies, research papers, and local knowledge.

Collaborate and share resources with your team, while having a secure personal space to save your own materials.

People can easily collaborate in a shared workspace, where they can view saved resources and comment to communicate with team members.

design system

How will the platform look, given that the design system is its backbone?

impact & Success Metrics

Currently

1000 research papers have been trained for the LLM.

Next

Increasing the data set to 1 million while improving accuracy.

Future

Developing and launching the tool, with lab affiliations.