Getting Started with SOMAK AI

Quick start guide to configure autonomous incident remediation.

Create Your Organization & Workspace

Every workspace in Somak AI is strictly isolated with independent error budgets, runbook patterns, and role-based permissions. Sign up at /signup or visit /onboarding/create-org.

Workspace Slug Architecture

Organization slugs are unique lowercase identifiers used in your dedicated inbound webhook endpoints (e.g. https://api.somak.ai/v1/webhook/ingest/your-slug). The user creating the workspace is automatically provisioned the Admin role.

Connect Sentry Telemetry

Somak AI ingests real-time crash reports via Sentry Webhooks. Navigate to Sentry → Settings → Integrations → Webhooks and paste your inbound URL:

Inbound Webhook Configuration

https://api.somak.ai/v1/webhook/ingest/<your-org-slug>

Connect AI Reasoning & Tavily Grounding

Somak AI uses a two-tier model architecture: Nemotron-3 30B for sub-10ms fingerprint triage, and Nemotron-3 550B MoE for AST syntax patch generation. Provide your API key in Settings → AI Engine or the Setup Wizard.

Zero-Hallucination Guarantee: Somak AI AST patches are verified in Sandbox sandboxes before canary promotion. LLM output that does not compile or pass tests is rejected immediately.

Configure Notification Channels

Receive live alerts on Slack, Microsoft Teams, and PagerDuty when an incident begins auto-mitigating. You can trigger a live test notification directly in the setup checklist to verify your webhook connectivity.

Invite Your SRE Team

Use the multi-email tag input to send team invitations with pre-configured roles (Admin, Operator, Viewer). Each invite generates a cryptographically secure 7-day token link.