STORY
Project charter and user stories
AUTHOR
ecurrencyhodlerecurrencyhodler
Joined 2023.07.01
PROJECT
Bitcoin-PALBitcoin-PAL
DATE
VOTES
sats
COMMENTS
NOSTR
View on Nostr ↗
TAGS
#activity#product-people

Project charter and user stories

After our call on 7/14, I wrote up a project charter and user stories to help clarify what exactly we were hoping to achieve based off of what was discussed. I find these both to be super helpful especially when working asynchronously.

Project charter is a one stop document for everything related to the project. It helps explain the project, typically includes a resource section with links, an appendix which documents decisions made throughout the project, and even metrics to help us know whether or not we've been successful. But I didn't think those were necessary for this hackathon. I may or may not regret this after!

User stories are helpful in that they lay out step by step the sequence of actions we want a user to take. This helps everyone (design/engineering/PM's) fill in the gaps wrt to their specific area of responsibility.

One funny thing I did for the hackathon was to write up a user story from an AI perspective. I felt like it was necessary because the AI itself was taking automated steps we needed to take into consideration. It also helped define the scope a bit of what we wanted to train the AI on.

*I reposted most of the doc below but you can find the project charter here. It also includes a link to a table with various team members, their skillsets, and timezones to help coordinate for meetings.

Project Charter

For the bolt.fun hackathon, hackers are challenged to revolutionize how we use Bitcoin and AI. Our project will contain two parts:

  1. A user-facing bitcoin education platform.

  2. An incentivized crowd-sourced training model for LLM’s using bitcoin. This will be the main focus of our project.

User Stories

There will be two user stories. First we will dive into the incentivized crowd-sourcing training model. Then we will do a very light demo of a chat-based bitcoin education platform.

1. Training LLM

In this scenario, an organization (FOSS or commercial) have already compiled and fed the LLM bitcoin data we consider to be relevant to the bitcoin education platform. This includes bitcoin core, bitcoin stack exchange, BIP’s, BOLT’s, and bitcoin op-tech. In light of this, the platform will be focused specifically on bitcoin developers. 

The organization has also funded the project with Bitcoin to incentivize submissions.

Contributor

  1. As a contributor, I want to submit a pdf or text of a content I think would be relevant to train the LLM and give it new info.

  2. As a contributor, I want to submit it to the LLM so that I can get bitcoin rewards.

    1. This will require me to provide a lightning or bitcoin address

  3. As a contributor, I want to verify that my submission has succeeded or failed with a message response.

  4. As a contributor, I want to receive bitcoin payment if my submission is successful.

LLM

  1. As an LLM, I want to receive data so that I can improve my training.

  2. As an LLM, I want to use various criteria to filter the data so that it doesn’t skew my current work.

    1. The criteria includes:

      1. Verifying acceptable sources such as: stack exchange (must have at least 50 votes and will accept the highest answer), bitcoin optech, BIP’s, LIP’s, and updated versions of bitcoin core.

      2. Must be above 50% relevancy with current database wrt to how related it is. (AI can provide % guesstimate)

  3. As an LLM, I want to pay out bitcoin to the submitter to reward them based on the quality of their submission.

2. Chat-based Bitcoin education platform

  1. As a user, I ask the bitcoin education platform a question about bitcoin so that I can learn about it by reading a text generated by an LLM.

  2. As a user, I want to know if the LLM isn’t familiar with the topic and then prompted to submit data to improve the LLM.

Demo Deliverables

I imagine us doing a video recording of both user flows during our presentation.

Flow #1

The demo will be a skeleton website. We basically want to demonstrate us asking the LLM a bitcoin question and it responding with that it doesn’t know. Then we feed it some data, it’s checked, and paid out. Then we ask the website again the same question and it provides an answer.

Key Dates

  1. July 31 - End of Hackathon

Stretch Goals

  1. As a contributor, I want to submit new categories to the database so that I can help increase the knowledge of the LLM if my submission fails. 

    1. I will be notified on how to submit another proposal through a pop-up within the bitcoin education chat platform.

      1. An example of new bitcoin content would be the pre-LN days. Or even ark.

  2. As an LLM, I want to pay out bitcoin to the submitter to reward them if they have a new category merged through a submitted PR.

Userflow #2

The second demo will show us going to the website and submitting data. Our LLM rejects it. Then we submit a PR which must be manually reviewed. Once it gets merged, the submitter gets paid out.