AI Application 5-Week Sprint

Project Build Sprint

Author: peterrr#4718

ELI5 (Explain it Like I’m 5) Project Value-Add

5-week sprint funding through the end of March to prove out long-term viability of LLMs project buildout with the end vision of unique fan/community generated analytics and insights for a future NBA franchise.

This is a modified resubmission of a previous proposal: https://snapshot.org/#/krausehouse.eth/proposal/0x4befc724eb167683a1d28b2be07f1aef6c13f4ca27388f1ac625081ba77e24a9

Requested Budget:

• 12,500 Krause payable at end of March

wallet: 0xBF6C4eE9441C774560b68D5ae34421d89C6F8787

Project Details

After feedback from various high-context members, I am focusing this sprint on one singular path with a clear vision. The end vision focuses on enabling a new implementation of data analytics in basketball and across sports, fan-enabled analytics. Large language models allow entities to turn language, conversation, and dialogue from unstructured data into actionable insights. As it relates to Krause House, with a unique data layer, we would be able to execute actionable insights for NBA teams such as an international / youth scouting network, individual and team scouting reports, and gameplay design including custom plays and strategies all enabled by fans and simple language.

While this serves as the grand vision, this proposal narrows focus on the first step to prove the viability of the project long term. The value is the data layer, meaning we must create an application that enables the collection of this text data to further turn into actionable insights. While there are many strategies, this proposed strategies focused on creating a unique application through LLMs which encourages participation and therefore data collection.

Project Stages: This proposal is only for the first Stage 1 but I want to outline long-term stages to show value in the first iteration, product building which enables the first data layer.

1. Conversational Basketball Application - A “jailbroken”, strong personality application that can cover all NBA topics, and knows all up-to-date stats, facts, and scores. Like DAN for Jerry’s by Jerry’s. Like Bill Simmons meets Stephen A Smith meets the Terminator, all with superior basketball knowledge. The intention is to push the boundaries of entertainment and LLM-based insights. This gives us the reps to start exploring LLMs for Krause House and enables us to begin sourcing fan opinions and content all while giving us some interesting media along the way. A conversational application that delivers multimedia content including conversational text, images, AI-generated images, and statistical graphics within conversations.
2. Building ML models for improved tagging and organization of data from data collected from efforts in stage 1. While this is possible through human intervention, this step would focus on building a system to create actional insights at scale. This allows us to transform fan and community-curated data at scale and transform it into structured data.
3. Build out of product, models, and incentive models to narrow focus in creating a scouting network, scouting reports, or gameplay design.

To reduce any worries about data collection, all data collection will be done anonymously.

Success metrics on the 5-week sprint

• Bi-weekly documentation of buildout
• Demos for interested members by end of week 2
• Working beta by end of week 5 on various messaging apps (telegram, discord, WhatsApp) and a custom browser-based platform.
• Full documentation to ensure no value is lost for future builders

Ultimately, a follow-up proposal after this 5 week sprint will allow the community to determine the quality of work and long-term value to the DAO. The next steps will be further outlined in the next proposal as we understand the implications of this sprint.