Polkadot’s OpenGov has matured into one of the most capable collective decision systems in crypto. Yet the information environment around it is fragmented. A typical voter’s journey crosses Polkassembly and SubSquare, dives into forum threads and bounty pages, peeks at GitHub activity, skims OGTracker and scans social channels like Reddit. Each step adds context. Each hop adds doubt:

• Where did a claim come from?
• Is a “new” proposal actually the same request across a different track?
• Did we fund something similar six months ago?

Clarys.AI is the governance intelligence layer designed to reduce that cognitive burden. We are building a provenance-first data backbone and a usable product, the Accountability Explorer, that unifies on-chain and off-chain governance signals, resolves entities across mechanisms, classifies duplication with clear definitions and reveals relationships visually. The data and interfaces are open by default: public GraphQL/REST APIs, daily CSV/Parquet exports and permissive licenses so any researcher or builder can reuse and extend the work without lock-in.

By integrating AI into OpenGov workflows, we can boost user productivity by up to 300%, eliminate over 120,000 hours of manual governance work, increase workload efficiency by 50% and save the treasury billions of dollars in unnecessary spending.

We listened to the community’s feedback on our prior referendum: Phase One must deliver something people can use. In response, this proposal centers not just on backend improvements but on a tangible, everyday tool for the ecosystem. Beta v1 ships an interactive explorer with related-proposal graph views, duplication flags, provenance popovers, filters and sortable tables, alongside open datasets and integration hooks for governance frontends. To make quality measurable and model-agnostic, we will publish a transparent evaluation harness: 150 canonical governance questions run nightly with accuracy, citation coverage, latency and cost dashboards, comparing open-source and commercial models under schema constraints.

The full proposal details the product, the data and architecture behind it, the implementation plan and a fully calculated budget for USD 130,000 to deliver Beta v1 in approximately twelve weeks.
Full proposal: https://docs.google.com/document/d/19en4NoDkrsRShj2_7L6w2KY35v2blkFAR2nAqM3nx5M/edit?usp=sharing

The first proof of concept version of Clarys.AI was developed as an MVP and its development was funded by Web3 Foundations Decentralized Future grant, you can read the intro article here. 

MVP Testing: https://clarys-ai-gamma.vercel.app (please note this is only a proof of concept version, bugs can occur and limited features are available at this stage).