# 緊急醫療物資供給的 matching tool

> I want to introduce you to Anuj and Nazz who are leading RxC Bangalore. They and a small team in India are hard at work building a tool to help match supply and demand for emergency medical supplies to meet the crisis going on right now in India. Within the last week they received $6M+ in funding from several prominent people in the cryptocurrency world.
>
> They are looking a lot at what g0v did in Taiwan and would love to connect with any folks from the community to learn from and possible partner with. Would you be open to speaking to them or know who to put them in touch with who might be able to help them get this right?
> -- Jennifer

> RxC Bangalore 詢問 g0v 裡面是否有人願意跟他們合作建置一個緊急醫療物資供給的 matching tool ，因應近期印度的醫療資源分配需求。 看起來經費已經到位，但需要有經驗的人能夠快速傳授經驗，請問這裡有人有興趣嗎？
> -- sylin

Response from RxC Bangalore:
> We're calling our demand/supply matching software "Covidbot". A working diagram of how it will work: https://raw.githubusercontent.com/cryptorelief/covidbot/main/working-diagram.png
> Our first release (ETA 3days) will plug into Telegram, with Twitter integration following soon. We've partnered with a community of almost 100k people (!) on Telegram, so our bot will start helping people and saving lives almost immediately after it is deployed.
> Although we're starting with Telegram commands (e.g. /find), the real potential of the bot will be unlocked when we figure out NLP/AI classification, so Telegram users will not even need to invoke our bot - it will silently reside in the group and read every request or supply information coming in, classify it, auto-match it with verified supply data and directly respond to the Telegram user. We're integrating our database with nationwide efforts that are aggregating healthcare supply data on a large scale and verifying it manually with the help of hundreds of volunteers (only ~3-5% of all supply data out there is actually useful since we're in a crisis and any verified supply gets used up within minutes).
> > It is important to note that we're currently working only with volunteers that joined the CryptoRelief community directly. Some of the code/modules being deployed are not on Github yet, but will be there soon.
> Besides building this aggregation-classification-matching-response system, we're currently putting in efforts towards building a basic NLP by ourselves. However, we are not experts at NLP or AI. If there are any experts or experienced professions in AI/NLP in your community, it would be amazing to have their help!
> We've also reached out to OpenAI privately but there has been no response so far. If you are able to get unlimited OpenAI or HuggingFace API access somehow, that would practically solve all our AI needs and accelerate our project significantly. If private efforts fail, we will do a public appeal on Twitter and your support there would be super useful as well!
> Apart from these, we are curious to learn about all your ideas, techniques, concerns, etc. regarding this project. Please do share all relevant past experience with similar things, as well as other things we can try.