Culpritv0 · 2026
Culprit / Alert Intelligence — V0

Alerts in with hostnames, emails, API keys.
Only tokens ever leave.

Most AI-powered ops tools ship your raw hostnames, emails, and API keys to OpenAI as training-accessible context. Culprit tokenizes at the edge — placeholders only leave your network. Logs, notifications, LLM prompts all see <HOST_a3f9>, never db-prod-7.acmecorp.internal.

0 PII in logs0 PII in LLMs0 PII in notifications
PAYLOADENCRYPTTO LLMAUTHORIZEDhost:db-prod-7user:alice@acmekey: sk-9f2…pgp_sym+ rlshost:<HOST_a3f9>user:<EMAIL_4d2c>key: <KEY_e8b1>host:db-prod-7user:alice@acmekey: sk-9f2…WEBHOOK/ INGRESSVAULT/ ENCRYPTEDAI/ CORRELATIONINCIDENT/ UI
Fig 01 — Tokenization pipeline
Memory / Adaptive RCA

Memory across incidents.

Most AI ops tools see one incident at a time, then forget. Culprit cites the top similar prior fixes on each service — the LLM reasons against your team's resolved cases, not from scratch. Mark an analysis wrong and it stops showing up next time.

Cites prior fixes on this serviceRecurring patterns surface as you goWrong analyses opt out of future context
01ALERT02RCA03RESOLVED04MEMORYcited by next RCAINCIDENT/ INCOMINGWITH/ PRIORSCLOSED/ + FEEDBACKINDEXED/ VECTOR
Fig 02 — Feedback loop

01 / 03 — Tokenization

Built to never see your data.

Your 3am database alert contains a real hostname, a real IP, maybe a real username in a stack trace. Most AI-powered ops tools ship that text straight to OpenAI. Culprit encrypts payloads at the edge and tokenizes them before any downstream system — log lines, notifications, LLM prompts see placeholders only. Rehydration is scoped to authenticated users with matching tenant scope.

How tokenization works

What Culprit sends

alert: auth_failure
host:  <HOST_a3f9>
email: <EMAIL_c44b>
ip:    <IP_b217>
auth:  Bearer <APIKEY_d8e9>
Same pattern, zero identity. AI correlates without ever seeing the real values.

02 / 03 — Correlation

AI clusters noise into incidents.

Correlation happens on tokenized events, so the model never sees raw customer names, IP addresses, or ticket contents. You see a single incident with a ranked root-cause summary — not a page of alert notifications.

03 / 03 — Aligned incentives

$49 per service, flat.

Most ops platforms charge per alert or per event — which trains you to sample, throttle, silence. Your AI gets a narrower signal, your correlations get worse, and your bill spikes exactly when you can least afford a distraction: during an outage.

Culprit charges $49 per service connection, flat. There's no per-event meter and no tier jump on heavy days — outage weeks cost the same as quiet ones. The more signal you send, the sharper your incident graph; we're aligned with you sending more, not less.

per service · per month

$49

Flat · No overage · No caps

04 — Frequently asked

Questions we answer on every call.

build b46460aupdated 2026-05-13no trackersno analyticsno third-party scripts

Get started · Culprit