Frequently Asked Questions

AlphaArc is a platform to run data-driven AI agents. When OpenAI and others trained their models they were only using Web2 data. We are here to bridge that gap. Our indexers and transformers analyze transactions, patterns, and market movements to provide clear, actionable information for LLMs and humans alike.

Our AI agents are specialized programs that continuously monitor blockchain activity. They receive tasks e.g. "find all transactions over 1M USD" and use large language models (LLM) to translate the intent of the input text into program calls and data queries to complete them. We augment agents with blockchain metadata data to provide real-time insights and provide their Web3 knowledge. Agents can be written with multiple frameworks e.g. Eliza, Langchain and languages (Rust, Python, JavaScript). For the LLM backend there are various options like OpenAI's GPTs, Anthropic's Claude, Meta's LLAMA or other self hosted models. Derived insights are delivered to you in a format that suits your needs: API call, email, tweet, Telegram message.. using integration plugins

Yes! Users can create and customize their own agents to focus on specific aspects of blockchain data, such as specific tokens, trading patterns, or market trends. Each agent can be tailored to your specific needs. We are still testing the platform but want to open it soon to the public.

Agents provide various insights including market trends, significant transactions, whale movements, trading opportunities, and protocol activities. These insights are delivered in real-time through your preferred format.

On the "Solana" data page you will see a heatmap of data that has successfully been indexed and transformed as well as some minimal sample data. There a numerous steps involved from loading raw block data to transforming it into a format agents understand. Currently, we are able to process around 1.7M transactions per second which is roughly 425x Solana's TPS. Our goal is to not only provide real time data but also give agents historical knowledge over significant events and actors.

Running algorithms at petabyte scale is expensive: https://x.com/tonykebot/status/1863832776535085117. We are fine tuning some steps and will be able to provide more data soon. Each dot represents many agent IQ points and our goal is to light up the whole heatmap like a Christmas tree.

Follow us on Twitter @AlphaArc4k for the latest updates and announcements or join our email waitlist to be the first to know about new features and releases.