NewHow the Agent Session Protocol builds trust between autonomous systems.
Product

Your events have a contract

Every event type is registered, versioned, and validated before delivery. One source of truth.

Events flow between organizations.

Nobody tracks what data is shared.

Nobody enforces schemas.

Nobody manages versions.

A breaking change propagates silently.

Consumers discover it in production.

Governance exists on paper.

Not in the infrastructure.

Until now.

One registry for every event type.

Every event is registered in the catalogue with its schema, owner, and subscribers. Validation blocks invalid payloads before delivery.

Evolve without breaking anyone.

Backward compatibility checks on every version change. Consumers upgrade at their own pace.

Step 1

Register Version

Define the schema, version, and access rules. The schema is the contract.

Step 2

Compatibility Check

Backward compatibility verified automatically. Breaking changes are blocked.

Step 3

Deliver

Subscribers declare their expected version. Negotiation is automatic.

Only the latest value per key.

Configuration, pricing tables, feature flags, and state sync. Compacted channels retain only the most recent value for each key. Storage stays constant regardless of update frequency.

Compacted Channel
pricing.standard
$29/mo
latest
pricing.standard
$19/mo
replaced
feature.dark_mode
enabled
latest
config.region
us-east-1
latest
Only latest values retained. Storage stays constant.

Years of history at cold storage prices.

Events migrate automatically from hot to warm to cold based on your policies. All tiers remain queryable.

Hot

Recent events, instant access. For real time workloads and the most frequent queries.

Warm

Older events, fast queries. Automatic cost reduction without sacrificing performance.

Cold

Years of history at the lowest cost. Remains queryable. Automatic migration by age policy.

Before and after Data Governance

Without Data Governance

  • No schema registry for events
  • Breaking changes propagate silently
  • No version negotiation between services
  • Events stored in one tier forever
  • Governance exists on paper only
  • Consumers discover breaking changes in production

With Data Governance

  • Searchable catalogue of every event type
  • Backward compatibility checked on every version
  • Automatic version negotiation per subscriber
  • Hot warm and cold tiers with automatic migration
  • Platform enforced governance rules
  • Invalid events rejected at publish time

npayload vs. building it yourself

FeaturenpayloadBuild it yourself
Event catalogue with searchWeeks of work
Schema validation on publishComplex to build
Backward compatibility checks
Compacted channelsMonths of work
Tiered storageMajor project
Platform enforced governance
Multi version negotiationComplex to build
Years of retentionMajor infrastructure

Frequently asked questions

What schema formats are supported?+
JSON Schema natively. Schemas are stored with version history, owner, and consumer list.
Can I have multiple versions of the same event type?+
Yes. Subscribers declare their expected version. npayload handles version negotiation automatically.
What is a compacted channel and when should I use one?+
Only the latest value per key is retained. Ideal for configuration, pricing tables, feature flags, and state sync.
How does tiered storage work?+
Define policies by event age. Events move from hot to warm to cold automatically. All tiers remain queryable.
We already define schemas in our code. Why do we need a registry?+
Schemas in code drift independently. The registry enforces compatibility at the platform level before changes go live.
What happens to existing events when I publish a new schema version?+
Nothing changes. Existing events stay in their original version. The new schema applies only to new publishes.
Can governance rules be different per team or environment?+
Yes. Scope rules per event type, channel, or environment. Relax rules in dev, enforce strictly in production.