The necessity of data provenance

In the current expansion of the AI agent ecosystem, the primary bottleneck is not just the reasoning capability of models, but the reliability of the data they ingest and produce. As autonomous agents begin to perform complex tasks—ranging from financial transactions to software development—the need for a verifiable audit trail becomes a requirement for any enterprise-grade deployment. Irys addresses this by positioning itself as a provenance layer rather than a standard storage provider. Founded by Josh Benaron and headquartered in London, the company began its life as Bundlr, a scaling solution for the Arweave network. It has since evolved into a standalone datachain designed to make data both permanent and traceable.

From Bundlr to a standalone datachain

The transition from Bundlr to Irys in late 2023 marked a shift in strategy. While Bundlr was primarily seen as a utility to make Arweave's permanent storage more accessible, Irys is a bet on the idea of "active data." This concept suggests that data should not just sit in a static archive but should be instantly available, programmable, and cryptographically linked to its creator. The Irys Datachain is optimized for high-throughput ingestion, allowing users to upload massive volumes of data with a guarantee of permanence. This is achieved through a decentralized network of nodes that verify and store information, ensuring that once a piece of data is committed to the chain, it cannot be altered or deleted.

Competitive standing and market role

Irys sits in a competitive space that includes established players like Filecoin and Arweave. However, it differentiates itself through its focus on developer experience and data throughput. Where Filecoin involves complex deal-making processes for storage, Irys offers a more straightforward, pay-per-byte model that feels more like using a traditional cloud service while retaining decentralized benefits. Its relationship with Arweave remains significant, as it still utilizes the network for long-term settlement, but its internal architecture allows for much faster transaction finality. This speed is what makes it attractive for applications that require frequent updates or real-time verification.

Infrastructure for the agentic era

For builders in the AI space, Irys is a tool for solving the "black box" problem. When an AI agent makes a decision, it can log its chain of thought and the specific data sources it used onto Irys. Because this log is permanent and verifiable, third parties can audit the agent’s performance with absolute certainty. This is particularly relevant for training data. If a model is trained on data stored on Irys, the provenance of that data—where it came from and when it was modified—is transparent. This transparency is a key differentiator for companies trying to avoid legal and ethical pitfalls associated with unverified training sets. By providing a permanent memory for the web, Irys is building the infrastructure that allows agents to operate with a shared, immutable history.