Summary
This month focused on completing our Q4 deliverables while preparing for 2025. We’re excited to continue our work on Horizon, making The Graph verifiable, and advancing our AI initiatives.
TAP and indexer-rs
Our November TAP efforts fell into three main areas. First, we helped Indexers transition from the old Indexer Service to our new one. We maintained a consistent presence at Indexer Office Hours and provided direct support for any technical issues. We improved our documentation based on user feedback and enhanced our library documentation to encourage contributions. This effort proved extremely successful—by the deprecation date, TAP was handling over 95% of network queries.
Second, we began planning TAP’s next major release, which is crucial for enabling Horizon. We collaborated with key stakeholders to identify user stories and develop a comprehensive plan. This plan is now being reviewed by core developers, Indexers, and members of The Graph’s Technical Advisory Board.
Third, we prepared for TAP’s evolution through a major code refactor. While these changes weren’t visible to users, they significantly improved the code’s architecture—making it more modular, testable, and extensible. These improvements position us well for adapting TAP for Horizon.
Next month, we’ll complete our preparatory work before beginning TAP’s adaptation. This includes finalizing the roadmap based on collected feedback and thoroughly testing to ensure we maintain stability while developing the next version.
Verifiable Extraction
We made significant progress in advancing Verifiable Extraction (VE) from research to production. We open-sourced the veemon monorepo, which features the vee library. vee provides a unified interface for verifiable extraction in Rust with pre-merge Ethereum blocks, supporting .dbin flat files, tools for handling protobuffer-defined Firehose block types, and cryptographic methods for verifying blocks.
In addition, we worked on creating proofs-of-concept (PoCs) for verifying post-merge data. We can verify all post-merge data up to 8192 blocks (an era) before chain head.
Next month, we’ll work on finishing up our final piece for post-merge verification, and we’ll start to integrate this functionality into our production library.
AI
This month, we wrapped up our user research around delegation, accompanied by a report on our findings and additional analysis of the subject from both a data and product perspective. Over the period of our research, we interviewed 26 users across five different user groups, gaining a depth of knowledge regarding the pain points users interacting with delegation face. We look forward to receiving feedback from our peers about our findings and to engaging with the ecosystem as we find the best path forward. From a natural language perspective, we have completed similar efforts and outlined a product and research roadmap that details our view on how to move forward with these technologies. We will continue to work closely with the Edge & Node team to continue to expand upon the body of user research that we have done in relation to these solutions and technologies. Additionally, members of our team actively participated in discussion around the intersection of blockchains and artificial intelligence at this most recent Devcon.
We are excited to be headed to NeurIPS to present on our work on synthetic data generation, which can be found here. We look forward to this upcoming year and the developments that are to come regarding The Graph and AI.