Platforms built a $300 billion industry on your listening history. Your cut is zero.

The music industry's discovery layer largely runs on fabricated data, except for what they know about you.

PR firms, managers, and labels engage in "trend simulation" to manufacture virality, deploying massive groups of tiny 'burner' accounts with the sole purpose of covertly placing songs in the background of posts in an effort to fabricate virality. Warner Records even offers this as a service for their signed artists. Firms like Floodify post thousands of videos every day across thousands of accounts in a coordinated effort to manufacture virality. But don't worry...

It gets worse.

Latest estimates from ISRC indicate that about 10% of music streams are fake. A federal lawsuit, filed in November 2025, paints an even worse picture. It alleges that Drake, who famously "started from the bottom", used money earned from gambling and promotion on Stake.us to hire bots to boost the play counts across platforms, distorting recommendation algorithms, and violating state consumer protection law.

And in 2026, artists now have to compete with low-effort AI generated music that's continuously flooding platforms, with services like Deezer seeing up to 39% of all uploads being AI and Apple music flagging over two billion AI generated tracks in 2025.

And the result?

We get the appearance of organic interest, but built by coordinated and fabricated personas. We get the concept of authentic discovery systematically hollowed out by content-cartels. For listeners, it means that your history, your behavioral patterns, built up over a decade or more, is not owned by you. Rather, it sits behind a closed doors, surfaced once a year (e.g. Spotify wrapped) yet always shared with third party advertisers. For artists, the relationship between you and your listeners is severed - you get the stats you get.

SOND3R

Sond3r is a local-first, privacy-preserving, agentic music client that runs on your own device, built on the Fangorn protocol. It works for you in the background, on your machine, not on a server. Your profile lives on your device, under your control, and goes where you go across contexts. It treats your taste as a first-class citizen rather than a resource to harvest, creating an organic, serendipitous, and shared human experience free from platform economics and misaligned incentives.

Sond3r gives you free and instant access to the web's largest public music library, no subscription or API keys needed. It also supports natively uploaded tracks with programmable access terms set by the artist.

Your taste, not their algorithm.

Services like Spotify collect data, feed it into their black-box algorithm optimized for retention, and use it to advertise to you. Sond3r builds a local model of your listening as you go. The more you use it, the better it understands your taste and what you actually want. And the best part: no ads. Ever.

Today, you just give it out, either for free (e.g. last.fm, owned by CBS) or worse, you pay for it (most streaming services). Sond3r enables a future where your taste, as curated by you, has value, and that value should flow to the party who built the taste. This profile can be exported on-demand, and eventually combined with other profiles to produce new kinds of aggregated curation abilities (road trip anyone?).

A shared brain, built by all of us

Users contribute to a collective picture of how music relates from real listening behavior across the whole network. Every listener can add to it, but nobody owns it. When you search or browse, you navigate a space built by everyone, but captured by nobody.

Context-Aware Discovery

An ambient agent runs alongside you as you listen. It reads your profile, queries the shared state, and empowers discovery, queuing, and recommendation, without sending your data to a third party advertiser or optimizing for attention retention. It enables things like..

• make a playlist for "poolside, eating ceviche, but my sunburn is getting bad"

• find the perfect song for "an intense car-chase" as it relates to

  

For Artists

Uploads and Access Control

The Artist Studio is in progress. At the heart of Sond3r is the Fangorn, a protocol that enables "intent-bound" data: data that can only be accessed by meeting some condition, like "you can read this if you paid me", the mathematical equivalent of "you must be this tall to ride". This means that access conditions for uploads are entirely up to the publisher.

1. Upload your music on your terms. Direct uploads IPFS and S3 are supported today. No agreements, contracts, or labels needed.

2. Programmable access control means your tracks are available as you see fit, be it free, be it one-time payments, pay-per-stream, token-gated, invite-only, time-limited, or any other condition

3. Get Paid instantly. Payments settle privately the moment a track is sold (within ~200ms). You are not subjected to a monthly payout cycle or minimum stream thresholds, and you don't participate in a shared pool for payouts.

If you are an artist interested in making your tracks natively available, reach out now. Your voice can be impactful in shaping the future of Sond3r for artists, and we want to hear from you!

Streaming services tell you if people listened. In the future, Sond3r will tell you where your music lives in the structure of human taste. This is beyond metadata, enabling real-time intelligence for artists and allowing them to better understand the relationship between them, their music, and their audience.

We're building in the open. Come find us.

Visit https://sond3r.com to learn more and to get involved.

We are actively seeking supporters who believe in this same vision.

First, some housekeeping:

• Fangorn was accepted into the Protocol Labs network!

• The team attended the Arbitrum Openhouse NYC event in early March.

• We built and entered the latest iteration of Fangorn as part of the PL Genesis Hackathon, here.

Fangorn Overview

We significantly revised and improved the core protocol this month, making it privacy preserving for purchasers, refining the gadgets framework, introducing structured data, and much more (details below). This makes encrypted data discoverable and understandable by a personal agentic curator, allowing you to discuss with your agent to find data (as opposed to surveillance + algorithmic recommendation). This is a new way to interact with others' data, bypassing centralized authorities, like platforms, and even allowing the agent to privately purchase data, without leaking behavioral signals.

Core Features

• Intent Bound Data: Fangorn allows data publishers to encrypt data behind 'intent' (specifically payment for now). The resulting ciphertext bundle is public and portable across any storage layer. It also offers an extensible gadgets framework that allows developers to implement bespoke unlock conditions.

• Privacy Preserving: For purchasers of data, Fangorn is fully privacy-preserving, ensuring no linkage between buyer, seller, resource, or price are encoded in storage.

• Schema ~ API: Fangorn relies on well-defined JSON schemas, registered onchain, that dictate structure, types, and which fields must be encrypted. Data can be published against a schema and consumed by an agent, with the SDK implicitly handling encryption of encrypted field types.

• Efficient Data Discovery: Data uploaded via Fangorn is indexed into a subgraph, making the datasources queryable by both agents and humans.

• Verifiable Data Provenance: Fangorn makes it trivial to trace who published data and when, with full transparency (and eventual accountability) for publishers.

How it Works

There are three major roles within Fangorn:

• The schema publisher defines schemas, uploads them to IPFS, and registers them within the schema registry contract.

• The data publisher builds schema-compliant datasets and registered them within the datasource registry (tracks storage roots and schema association) and the settlement registry (tracks price and enables private purchases).

• The data consumer uses the Fangorn agent to discover published data via a subgraph interacted with through the Fangorn MCP server. The data consumer's agent then uses the x402f rails to purchase data access and decrypt it locally on behalf of the consumer.

Data Discovery and Agentic Curation

A unique capability of Fangorn is that it makes data discoverable to both humans and agents, allowing for decentralized applications that 'feel' like web2, and also understandable by agentic systems. That is, we built and deployed subgraphs that monitor the Fangorn registries using a generic, schema-based data mapping. This means that our subgraphs can store any type of structured data in a single subgraph.

In other words, Fangorn allows for new schemas to be published, new data to be published against them, and then an agent can 'simply' find , purchase, and decrypt data as needed.

The Fangorn MCP server, live on cloudrun, lets agents interact with subgraphs for data discovery and UI rendering.

You can check out the Fangorn Agent (in the demo above) to experience the fully-functional, purely agentic, end-to-end flow of discovering, purchasing, and consuming data from a chat context. With the Fangorn agent, using your own agent, it goes out, finds data based on your intent, pays for it privately using zero knowledge proofs, and delivers the results to you, never relying on a centralized authority in the middle.

Fangorn SDK

Github

We just released the latest version of our SDK here https://www.npmjs.com/package/@fangorn-network/sdk

This introduces a structured data approach to Fangorn, with the idea being you register a schema, publish secret data against it, register access control conditions onchain, and make it available to others (or to agents).

It also introduces three 'registry' contracts:

• Schema Registry: A basic on-chain registry of storage roots (CIDs) of schemas stored in IPFS.

• Datasource Registry: Allows for data to be published against a schema. This is also the core mechanism that makes data discoverable, with events emitted from the contract used to trigger indexing in a subgraph.

• Settlement Registry: Enables programmable and verifiable access control conditions to be stored. It uses Semaphore to ensures no linkage between buyer-seller-resource-price is visible onchain.

  • Each entry in the datasource registry also gets registered in the settlement registry, creating a 'resource' - essentially a fresh semaphore group and corresponding price for registration into the group.

  • Settlement for access to data relies on a two-phase register/claim flow, where first a transaction must be confirmed that pays the owner the specific amount (using ERC-3009), then the given identity is registered in the semaphore group.

  • The second phase, claim, takes a groth16 proof of membership into the semaphore group and then emits the settlement event. This also can trigger an additional hook contract when proofs are valid, allowing for custom behavior (e.g. minting an NFT, ownership of which is used as the access control condition).

x402f

Github

x402f is a payment and access-control layer for encrypted data. It extends the x402 protocol with a register/claim model backed by Fangorn and Semaphore ZK proofs. Buyers get private purchases, private retrieval, and no on-chain linkage between their identity and the resource they accessed. Sellers gain easy and verifiable programmability for their access condition, able to configure price or other conditions (like NFT ownership) for gating decryption. In addition, no party is required to support additional infrastructure, requiring no server to release data.

• The semi-trusted facilitator never holds plaintext, cannot forge proofs, how it can technically withhold data delivery after settlement - a point of trust we will eliminate in the near future.

• We published two new packages:

  • fetch https://www.npmjs.com/package/@fangorn-network/fetch

  • facilitator https://www.npmjs.com/package/@fangorn-network/facilitator

Next Steps

• We are in early-stage chats with projects that have a large amount of data that may be made available via Fangorn, more to come soon!

• The tooling is still maturing, but is live on testnet! Please reach out if you have any questions, concerns, want to see a demo, etc.

Contact

• Live demo: https://vault-demo.fangorn.network

• Github: https://github.com/fangorn-network

• Website https://fangorn.network

• Join our discord! https://discord.gg/JDj8RdCVyU

When information is nearly free, value is not derived through controlling access, but in how information is curated, shared, and acted on. Today's centralized control and curation implies platforms are arbiters of shared-meaning, with a primary goal to maximize profit, not social welfare. Allowing AI agents to infer intent is fickle, often relying on invasive surveillance and with massive variance depending on models and contexts. The future web, if it is to be fair and open, requires transforming data form a platform-controlled resource to a programmable commons.

{ proof > permission }

The concept behind Fangorn is simple: Data access should be enforced by cryptographic proof, not platform permissions.

Instead of trusting platforms to provide permission to access your data, we're building access control based on proof. Fangorn enables secrets that enforce their own access rules. We call this intent-bound data.

Intent-bound data is encrypted information that can only be decrypted by someone who mathematically proves they meet some condition without revealing what they're accessing, and without relying on any trusted intermediaries.

Fangorn is programmable secrets infrastructure for the Agentic web: encrypted information that enforces its own access rules, consumed by humans and agents alike.

At a high level, Fangorn combines:

• client-side encryption

• zero-knowledge proofs for access control

• trustless, on-chain verification

• decentralized threshold encryption and storage

An Example

Consider one of the most prevalent, yet simplest, use cases: password-protected data.

Today, this generally requires trusting a server:

• to store and verify the password (or hash of it).

• to determine if and when you should get access, based on opaque platform policies.

• that can log, leak, revoke, and misuse your data.

With Fangorn:

• Data is always encrypted client-side; no server ever sees the plaintext.

• The password never leaves your device.

• Encrypted data exists in decentralized storage, making it instantly portable.

• No intermediary can access the secret or decrypt anything.

• No logs to leak, no unfair revocations.

If you can prove you know the password, you can decrypt the data. Otherwise, it remains locked.

Why This Matters

Today, sharing data on a platform is subject to the platform’s policies, incentives, and mistakes. Access can be revoked, personal data can be leaked, and “ownership” is really a temporary license. Fangorn eliminates any need to trust platforms with enforcement, but the problem we face isn't storage. Decentralized storage already exists, but adoption lags since, without privacy-preserving access control, using it means publicizing everything.

Fangorn fills this gap by separating storage from access, proof from identity, and stripping enforcement powers from platforms. It unlocks new classes of applications where data is shareable without surrendering control, is consumable by AI agents without surveillance, and whose access is governed by math instead of terms of service.

What You Can Build

• Agentic Dark Pools

  AI agents executing trades based on private signals that only decrypt under specific market conditions (e.g. RSI thresholds).

• ZK-Gated Content

  Content that only releases by meeting creator-controlled rules, independent of platform policies.

  • Timelocked Data

    Secrets that becomes public after a predetermined time (e.g. a Deadman's switch)

  • Token-Gated Assets

    Access control determined on-chain through token ownership (e.g. with perpetual royalties enforced by math, not platforms).

• Peer-to-Peer games with hidden state

  Including sealed-bid auctions and rock-paper-scissors, and other trustless games requiring private inputs.

• Cross-organizational data sharing and collaborative AI Training

  Many parties can encrypt private data to a 'data lake' that only decrypts for a specific TEE, or model trainer, or data clean room.

These are early ideas and examples, not an exhaustive list

Our Journey: Where we are now

Fangorn emerged out of several years of research and prototyping at Ideal Labs, where we learned what it takes to move from cryptographic theory to production ready systems.

• In December 2025, we won first place in the Polkadot 2.0 hackathon for a proof-of-concept threshold encryption network. The project demonstrated a novel approach to decentralized key management and programmable access conditions, and the win validated that we can design and implement complex cryptographic systems. However, the experience also revealed a harsh reality: engineering bespoke, novel cryptography is slow, risky, and often orthogonal to real-world usability.

• In January 2026, we began rearchitecting the system for a pragmatic, production ready deployment. Rather than reinventing core primitives, we focused on building on battle-tested infrastructure and modern zero-knowledge tooling, with privacy, composability, and developer experience as a fundamental focus from day one.

What's Next?

The first supported access-control condition is deliberately simple: password-protected data that does not leak the content being accessed, not to anyone. This forms the foundation of a broader, extensible gadgets framework, enabling data to be encrypted behind a composition of various gadgets, like: timelocks, membership checks, token ownership, and additional conditions under development. Fangorn aims to build a modular and extensible zk-gadgets SDK and framework, where new zk-access-control-conditions can be developed and registered, and verified trustlessly on-chain.

Join us as we build!

If you're reading then, you're early. Fangorn's ideas are ambitious, and the first tools are almost ready. If you believe the future of the web should be built on proof, not permission, we'd like you to build with us.

💬 Discord

🌲https://fangorn.network

💻 https://github.com/fangorn-network