Status

Status

What runs today and what does not. This page lists the services that exist, not an uptime figure we have not earned the right to publish.

What this page is and is not

This is a service inventory with current operational state, reviewed against the running system. It is not a live status feed: there is no automated probe behind it, no incident history, and no uptime percentage. A hand-maintained page that says so is more useful than a dashboard of numbers nothing measures.

A machine-readable status and freshness endpoint is planned alongside the Data API. Until then, last reviewed: 2026-08-31.

ServiceStateCadenceNote
Asset registryOperationalcontinuousCanonical identities and provider mapping.
Market observationsOperationalhourlyHourly snapshot written per asset.
Social observationsOperationalhourlySame grid as market.
SentimentOperationalhourlyWritten with the market and social snapshot.
News flowOperationalhourlyRolling window.
Narrative snapshotsOperationalhourlyTopic clustering over news flow.
Ecosystem seriesOperationaldailyCurated asset set.
Developer activity seriesOperationaldailyCurated asset set.
Historical storeOperationalappend-onlyImmutable observations.
Public Data APIIn developmentNo public endpoint exists.
WebhooksIn developmentNot implemented.
Dataset distributionIn developmentLicensing review, see datasets.
MCP / agent toolsIn developmentNot implemented.

Data source health

Every data category depends on an upstream provider, and when one degrades the category degrades with it. What we publish is the state of our own service for each category — operational, delayed, or degraded — without naming the providers behind them.

That omission is deliberate. Provider relationships carry commercial terms, and publishing a per-provider health board would expose contractual detail and give a precise map of which single upstream failure takes which category down. The category-level state is what a consumer needs; the attribution is not.

How outages are handled

When an upstream source fails, the affected job aborts rather than writing a partial snapshot, and the hour stays empty. No value is fabricated to fill the gap. That means an outage is visible in the data as a gap, which is the most honest status signal available and the one that reaches every consumer without them having to check this page.

The full failure-mode table is on data quality.

Reporting a problem

If a series looks wrong — a gap that should not be there, a value that cannot be right, an asset resolving to the wrong record — tell us: contact@neob.ai. Include the asset, the timestamp and what you expected. Resolution errors in particular are worth reporting, because they are the failures that propagate furthest.