v2 · narrative + agent economy

Knowledge spreads.
Rewards flow back.

AgriDiffuse models how agricultural practices move through a network of experts and farmers — like an epidemiological spread, but for ideas. Same graph, three views: structural, narrative, economic.

Open the graph Run a round →

Neo4j · Featherless.ai · Masumi.network · Lovable

The graph carries the story.

One network, three lenses. Each round emits structured events that flow through three layers — what happened, what it means, and who got paid.

Structural

Watch the spread.

Three node types, four edge types. Practices move along ADVISES and KNOWS edges; trust weight decides who picks them up.

Narrative

Hear the story.

Featherless.ai turns each round into a paragraph. One LLM call per round — structured graph events become plain language.

Economic

See the value.

Masumi.network gives experts agent identities. Every adoption with a traceable expert ancestor pays a reward — direct adoption pays 1×, each hop halves it.

Spread

Trust-weighted propagation. A farmer's adoption probability = expert credibility × farmer base rate × edge trust.

Trace

Every new adoption walks back to its closest expert ancestor along RECOMMENDS · ADOPTED · KNOWS*.

Reward

Masumi pays the expert. Direct adoption = 1×. Each peer hop halves the reward, mirroring influence decay.

Narrate

Featherless gets a structured event log — returns one paragraph for the simulator's narrative box.

Ready to see knowledge move?