Your on-call engineer gets answers, not just alerts
Wachd receives alerts from Grafana, Datadog, and Prometheus and tells you WHY it fired — not just that it fired. Self-hosted on Kubernetes. Your data never leaves your cluster.
3am. Your phone rings. You open the alert.
It says: “HighErrorRate firing.”
You still don't know where to start.
Wachd already ran the diagnosis while the alert was routing to you.
How it works
Alert fires in Grafana, Datadog, or Prometheus
Your monitoring tool sends a webhook to Wachd. Signature validated, event queued in Redis.
Wachd collects context automatically
Last 10 commits from GitHub. 30 min of error logs from Loki or Datadog. Metric history around the alert window.
AI analyzes the causal timeline
PII stripped first. Then Ollama, vLLM, or any OpenAI-compatible model in-cluster — or Claude, OpenAI, or Gemini — builds a plain-English root cause with a suggested action.
On-call gets the answer, not just the page
SMS, voice call, email, and Slack — in the order you configured. Escalation if not acknowledged. Cancel when resolved.
Everything your on-call team needs
AI root cause analysis
Correlates commits, logs, and metrics into a 2-sentence answer. Powered by Ollama, Claude, OpenAI, or Gemini — your choice.
Flexible on-call schedules
Primary, secondary, and custom rotation layers. Time-window restrictions (business hours, nights, weekends). Self-service overrides without admin.
Per-user notification rules
Email immediately. SMS at the same time. Voice call after 10 minutes if unacknowledged. Each engineer sets their own preference.
Air-gapped ready
Run Ollama, vLLM, or any OpenAI-compatible model in-cluster with no outbound API calls. Fully functional in isolated VPCs and classified environments.
SSO + local users
Entra ID, Okta, and Google Workspace group mappings. Local accounts always available as break-glass. Both are first-class.
PII sanitisation
Emails, IPs, API keys, and account IDs are stripped before any data touches the AI backend. Non-negotiable, runs synchronously.
CVE breach intelligence
Daily scan against NVD, OSV, and CISA KEV. Cross-referenced against your team's actual dependency inventory. Auto-opens fix PRs.
Multi-tenant isolation
Every team is fully isolated. Team A cannot see Team B's incidents, schedules, or data sources. Enforced at every query.
Self-hosted on Kubernetes
Helm chart with sane defaults. Supports external RDS/ElastiCache or in-cluster Postgres/Redis. Apache 2.0, no vendor lock-in.
Read-only by design
Wachd connects to your infrastructure to understand it — not to change it. It cannot deploy, restart, scale, or modify anything in your environment. Ever.
contents:read scope only. Wachd reads commits and lock files. It cannot push, open PRs, or modify any repository.
Query-only access to Loki, Datadog, Prometheus, and Grafana. No write operations, no configuration changes.
Wachd runs inside your cluster and makes outbound API calls only. No inbound access to your services beyond the webhook endpoint.
Your AI runs inside your cluster
Wachd ships with first-class Ollama support. Run Llama, Mistral, or any open model in-cluster — incident data never reaches an external API. One config line to switch backends.
- ✓GDPR and data residency — Incident logs, commit messages, and error traces stay inside your VPC. No third-party data processor agreement needed.
- ✓No model licensing risk — You control the weights. Licensing changes at Anthropic or OpenAI don't affect your on-call workflow.
- ✓Air-gapped environments — Fully functional with zero outbound API calls. Works in classified networks, isolated VPCs, and disconnected clusters.
# values.yaml — switch AI backend with one line
ai:
backend: ollama
model: llama3.2
ollamaEndpoint: http://ollama:11434
# vLLM, TGI, or any OpenAI-compatible endpoint:
ollamaEndpoint: http://my-vllm-service:8000
# or switch to cloud AI:
backend: claude # ollama | vllm | claude | openai | gemini
Works with your existing stack
Pricing
Start free and self-hosted. Pay only when you need more.
Open Source
Free
Self-hosted, unlimited teams
- ✓ Core AI alert analysis (Ollama)
- ✓ On-call scheduling + escalation
- ✓ Per-user notification rules
- ✓ Slack, email, SMS, voice
- ✓ CVE breach intelligence
- ✓ Unlimited teams and users
- ✓ Apache 2.0 license
SMB
$99/mo
For growing engineering teams
- ✓ Everything in Open Source
- ✓ Cloud AI: Claude, OpenAI, Gemini
- ✓ 30-day incident history exports
- ✓ Priority email support (next business day)
- ✓ Commercial license
Enterprise
Custom
For regulated and large-scale orgs
- ✓ Everything in SMB
- ✓ SSO: Entra ID, Okta, Google Workspace
- ✓ Compliance reports (SOC2, ISO 27001)
- ✓ Full audit logs
- ✓ SLA + dedicated support
- ✓ Professional services
Email sales@wachd.io to join the waitlist
Lifetime SMB license — free, forever
We're giving 10 engineering teams a lifetime Wachd SMB license at no cost. No monthly fee. No expiry. In exchange: use it seriously and tell us what's broken.
✦ What you get
- ✓ Full SMB license — lifetime, no subscription
- ✓ Cloud AI: Claude, OpenAI, Gemini
- ✓ 30-day incident history exports
- ✓ Priority email support
- ✓ Founder badge on wachd.io
- ✓ Direct line to the team — shape the roadmap
↩ What we ask
- → Deploy Wachd in a real environment — not a sandbox
- → Tell us what breaks, what's confusing, what's missing
- → A short conversation with the team (~30 min) if you're willing
- → That's it — no NDA, no contract, no catch
Or email sales@wachd.io with your company name
Ready to switch?
Deploy in under 30 minutes on any Kubernetes cluster. No vendor lock-in. No data leaves your infrastructure.
Deploy free on Kubernetes →Apache 2.0 · Self-hosted · No account required