Nerve sits above the monitoring stack you already have. It maps reliability signals to customer accounts, compares each account with its peers and baseline, joins support, change, usage and revenue context, and opens one evidence-backed customer incident.
If all you need is a 403, 500, timeout or latency graph, you do not need Nerve.
Nerve is the layer after detection: customer identity, isolation, business impact, evidence correlation, prioritization and recovery verification.
Useful technical evidence. The hard part is deciding who is actually hurt and what deserves attention first.
One affected account, one evidence trail, one incident to investigate while the global service still looks healthy.
Datadog, Sentry, New Relic and your own logs are already good at collecting technical signals. Nerve does not ask you to replace them.
The missing unit is the customer account. Nerve asks whether one tenant is deteriorating while peers remain healthy, then attaches the context an operator needs to decide what to do.
Resolve telemetry and external identifiers to the customer that experienced the signal.
Separate a customer-specific degradation from a global incident by comparing account and peer behavior.
Bring reliability, support, recent changes, product usage and revenue context into one bounded investigation.
Rank current customer-risk indicators and show latest known recurring revenue exposed while an account is at risk.
Surface concrete investigation actions grounded in the same evidence instead of a disconnected alert pile.
Do not close the loop on a status change alone. Require fresh healthy telemetry before recovery is verified.
This is the workflow Nerve implements. Existing tools produce signals; Nerve turns those signals into a customer-level operational decision.
Structured telemetry, support tickets, changes, usage and revenue enter through the ingestion API, SDK or configured connectors.
403 · /salesforce/sync · 821ms
Nerve maps provider-specific identifiers to one account instead of treating each source as a separate universe.
tenant_1288 → Northstar
Nerve computes the account's success, latency and retry behavior, compares it with recent history and workspace peers, and checks usage deterioration.
25% account · >98% global
A material customer anomaly becomes a deduplicated incident with the affected account, severity, confidence, evidence and probable cause.
Northstar · Salesforce Sync
Support urgency, recent changes, recurring revenue and product usage sit beside the technical evidence so the team can prioritize the right account.
$84K ARR · 2 tickets · cfg_v18
After mitigation, Nerve looks for a fresh recovery window. The incident closes only when the recovery conditions are actually met.
≥95% success · 0×403 · no new ticket
The Northstar scenario below is deterministic synthetic demo data from the Nerve platform. It is not customer traction or a production benchmark.
Nerve does not magically see an uninstrumented visual bug. It reasons over the structured signals your systems expose and the account context you provide.
Success and failure rates for one account versus its own recent baseline and workspace peers.
Which endpoints and customers carry a disproportionate share of recent failures.
Whether one account's latency or retry behavior is moving materially from its baseline.
Bounded changes in instrumented product-event volume from sources such as PostHog or Segment.
Recent tickets and priority signals correlated to the same customer and time window.
Recent account or global changes placed beside the deterioration as correlation, not automatic proof of causation.
Latest known recurring revenue shown when an account is currently at risk, without pretending it is churn probability.
Fresh post-mitigation telemetry confirms whether the customer actually recovered before the incident closes.
Connector modules and fixture contracts exist in the design-partner beta. Live provider credentials and external OAuth/webhook delivery are separate deployment validation steps.
The detector computes customer-level features and risk deterministically. Optional Azure OpenAI synthesis receives a bounded, sanitized evidence set and can produce concise summaries, risk drivers and recommended actions.
Raw signals are normalized into account-level features.
Deterministic logic decides whether material deterioration exists.
AI may only cite evidence IDs Nerve supplied.
If AI fails or returns unsupported evidence, Nerve falls back to deterministic intelligence.
The platform keeps technical and business context together without presenting a risk score as a calibrated churn probability.
The product is deliberately narrower than the pitch might sound. These boundaries are important.
Keep Sentry, Datadog, New Relic and your logs. Nerve consumes or receives structured signals from that layer.
If a UI problem produces no telemetry, product event, support signal or other instrumentation, Nerve cannot infer it from nothing.
Deterministic features and detection stay authoritative. AI output is bounded and evidence-grounded.
The current beta recommends and coordinates investigation. It does not claim safe autonomous production remediation.
We are looking for technical B2B SaaS teams willing to test the customer-impact layer against real reliability incidents and tell us whether it changes what they prioritize.