Kimacast

Discoverable AI services that generate, distribute & monetize licensed micro-content on social media

Kimacast

Created At

ETHGlobal Lisbon 2026

Project Description

Publisher Agents is an opt-in syndication platform that lets authors and publishers transform their existing content into discoverable, monetizable AI services.

A creator signs in through Privy, connects an authorized source such as an owned website, RSS feed, or uploaded file, and configures the agent’s tone, supported formats, content policy, approval rules, and payout address.

The agent uses 0G Compute to understand the source and generate source-grounded formats such as social posts, newsletter briefs, threads, and machine-readable citation bundles. 0G Storage preserves source snapshots, author preferences, agent memory, generated outputs, and provenance references.

Each agent receives a programmatically created ENS subname, such as alice-tech.publisher-agents.eth. The name becomes the agent’s persistent identity and discovery point, resolving to its wallet, capabilities, profile, reputation data, and interaction endpoint.

A downstream consumer—another AI agent, newsletter, aggregator, research bot, or publishing service—can request a licensed syndication bundle from the agent’s API. The API responds with HTTP 402 Payment Required; the consumer authorizes an x402 payment settled in HBAR on Hedera, and the requested content is unlocked after settlement.

Every successful delivery contributes verifiable signals such as paid purchases, unique buyers, delivery success, refunds, revenue, and provenance. These signals update the agent’s reputation profile, which is discoverable through its ENS identity.

As a secondary feature, creators can launch follower-guided prefunding campaigns. Supporters contribute small payments toward proposed topics, helping creators validate demand and fund the next piece of content before producing it

DEMO JOURNEY

  1. Create a publisher agent

The creator signs in with Privy and provides:

authorized source → content formats → publishing preferences → content policy → payout address

  1. Provision its identity

The platform programmatically creates:

alice-tech.publisher-agents.eth

The frontend demonstrates that the name resolves dynamically to the agent’s profile and endpoint.

  1. Ingest an authorized source

The agent retrieves one article or file and stores a source snapshot and provenance reference through the 0G-backed pipeline.

  1. Generate micro-content

0G Compute produces:

an X-style post; a three-item thread; a newsletter or LinkedIn-style summary; a machine-readable citation bundle.

The UI displays the model or provider information and the resulting 0G Storage references.

  1. Request premium content

A second agent requests:

GET /agents/alice-tech.publisher-agents.eth/content/123

The API responds:

HTTP/1.1 402 Payment Required 6. Settle on Hedera

The buyer signs the value transfer. The facilitator sponsors the network fee and settles the transaction on Hedera Testnet.

The UI shows the transaction receipt before unlocking the bundle.

  1. Update reputation

The dashboard refreshes:

Paid deliveries 14 Unique buyers 6 Delivery success 100% Revenue 18.42 HBAR Reputation 82 → 84

The agent’s ENS profile points consumers to the updated reputation evidence.

WORKFLOWS :

Content creator │ │ SET content-generation-settings ▼ Publisher Agent │ │ Generate ENS records and submit premium-content ▼ Social media API (Approved publication)

Consumer Agent │ │ GET premium-content ▼ Publisher Agent API │ ├── HTTP 402 Payment Required │ ▼ x402 payment │ ▼ Hedera Testnet settlement │ ▼ Content unlocked │ ▼ Author revenue + reputation updated

Consumer Agent │ │ Publish content-preference-polling ▼ Publisher Agent API │ ├── HTTP 402 Payment Required │ ▼ x402 payment │ ▼ Hedera Testnet settlement │ ▼ Author pre-funding updated │ ├── Submit new content ▼ Publisher Agent API │ │ Generate ENS records and submit collectively-supported-content ▼ Social media API (Approved publication available only to supporters)

How it's Made

The application is organized into four main layers: creator onboarding, AI generation and storage, agent identity, and machine-to-machine payments.

Creator onboarding

Creators authenticate with Privy, which provisions an embedded EVM wallet. From the dashboard, a creator connects an authorized URL, RSS feed, or uploaded document and defines the agent’s tone, target audience, output formats, content restrictions, approval settings, and payout address.

For the hackathon prototype, we deliberately avoid training a separate model for every creator. Each agent uses:

a shared model + per-author retrieval corpus + author profile + content policy + persistent memory

This keeps agent creation fast while still producing author-specific behavior.

0G Compute and Storage

0G Compute is the inference layer. It processes the source content together with the author profile and generates structured outputs such as an X thread, newsletter summary, short social post, and citation bundle.

0G Storage is used for source snapshots, author configuration, generation metadata, approved outputs, provenance references, and evolving agent memory. A generation record links the output to its source identifier, prompt version, agent identity, and storage reference.

ENS identity and discovery

When an agent is created, the platform programmatically assigns it an ENS subname such as:

alice-tech.publisher-agents.eth

The name resolves to the agent wallet and stores discovery metadata through ENS text records. We use the ENS agent-record model to expose an agent-context and a web or API endpoint. This lets another application discover how to interact with the agent using only its ENS name.

ENS is the identity and discovery root, but it does not self-assert reputation. The reputation shown on the agent profile is calculated from observable events such as:

successful paid deliveries; unique purchasers; failed or refunded deliveries; cumulative revenue; provenance records; repeat purchasers. Hedera and x402 payments

Premium content is exposed through an x402-protected endpoint. When a client requests a licensed bundle, the API returns HTTP 402 Payment Required with the amount, accepted asset, destination, and facilitator information.

The consumer signs the value transfer, while the Hedera x402 facilitator acts as the transaction fee payer, verifies the payment, submits it to Hedera, and returns the settlement receipt. After confirmation, the API releases the content bundle and records the delivery against the agent’s reputation profile.

For the MVP, payments are priced in native HBAR. This avoids introducing an additional HTS-token association flow and keeps the demonstration focused on one discrete request, one payment, and one content response.

Prefunding extension

The prefunding flow reuses the same payment infrastructure. A creator publishes several proposed topics, supporters make small x402 contributions, and Hedera receipts update each campaign total. When the target or deadline is reached, the selected topic moves into the 0G generation workflow. Supporter receipts can later be used to provide early access before the creator approves a public release.

Technical Considerations: The frontend provides Privy-based authentication, an embedded EVM wallet, creator onboarding, agent configuration, generation review, sales analytics, and prefunding campaign management.

When a creator provisions an agent, the backend creates an author profile containing the authorized source, supported formats, tone, policy constraints, payout destination, and approval settings. Instead of fine-tuning one model per creator, the prototype combines a shared inference model with a creator-specific retrieval corpus and persistent configuration.

The generation pipeline retrieves the authorized source and profile, invokes 0G Compute, validates the structured response, and uploads the source snapshot, output, prompt version, and provenance metadata to 0G Storage. The resulting storage identifiers are attached to the agent profile and generated bundle.

The identity service creates an ENS subname for every agent. We publish an agent-context record describing the agent’s capabilities and an agent-endpoint[web] record pointing to its public profile or API. A client can therefore start with only an ENS name and discover the agent address, supported content formats, provenance references, payment endpoint, and reputation profile.

Premium bundles are protected by an x402 middleware layer. An unpaid request returns 402 Payment Required. The consumer creates and signs the Hedera transfer authorization, and the facilitator verifies and settles the transaction as fee payer. Following settlement, the resource server returns the licensed bundle and records the transaction ID, buyer, delivery status, and revenue.

The reputation service aggregates payment and delivery events rather than trusting a creator-editable ENS value. ENS points to the reputation profile, while Hedera settlement receipts and stored provenance provide the supporting evidence.

background image mobile

Join the mailing list

Get the latest news and updates