The story behind building Hemera

The idea originated from the learning process by a team of AI & ML experts serving hundreds of Dapps and Infra projects' growth teams for 2+ years.

Over 2 years ago, when Hemera team (fka. w3w.ai) first came to build in web3, we were amazed by the potential of crypto data --- an ever-growing data network where each piece of the record owned by their rightful owners (the end users) in a decentralized and trustless way. Instantly we were picturing the future of internet where web3's equivalence of Facebook, TikTok, Amazon, OpenAI and Google thrive, the massive user-owned data would allow 3rd party developer to build experiences that the internet has never seen before.

Then two main questions quickly showed up in our heads:

1. Are there enough demands in the industry for on-chain data now?

2. Any opportunities to build a durable project on top of a data set anyone can access?

For the first question regarding demands, the answer is a resounding YES. The team very quickly acquired close to 1000 projects as customers using our first application (a web3 CRM) production to help community growth and airdrops. The projects had a wide range of backgrounds, including Dapps like GameFi, NFTs, SocialFi, and infrastructures like blockchains and protocols ---- all of them needed some way to differentiate quality users on-chain from sybils, bots and fake accounts.

Answers for the second question weren't obvious until we realized building a new infrastructure for the industry was in fact most needed: while we were trying to build the CRM application to help our customers, we in fact ended up spending up to 90% of the engineering resources building our internal Machine Learning (ML) & data infrastructures from scratch (starting from pulling raw transaction data from RPC nodes for each blockchain we were supporting). Thus we decided to turn our learnings in the infrastructure layer to a novel indexing protocol to help all web3 developers abstract out complexities involved with processing and computing semantics data out of on-chain transactions, hence the Hemera Protocol.

Team background: Hemera's founding team built enterprise software powered by ML & Deep Learning (DL) in the Silicon Valley, and sold the it to some of the largest enterprises in both US and Asia to help them understand user behaviors: detect sybil attackers and identify potential big spenders (whales). This unique experience in web2 gave the Hemera team lots of insights in how to build large scale AI & data systems to support various ML & AI modeling use cases.