Built for support teams. AI that thinks first.
RAUM runs every ticket through four steps — Understand the question, Plan the resolution, Act on your connected tools, and Reply in your brand voice. No single-shot guesses.
How RAUM works
- Understand
Reads the ticket, figures out what the customer actually wants, and pulls in context from Zendesk tags, custom fields, and the full conversation history.
- Plan
Searches your decision trees and knowledge base, picks the right tools, and maps out a step-by-step path to resolution before taking any action.
- Act
Runs the plan — calls into your connected tools, fetches live data, and takes action automatically when you've enabled it.
- Reply
Writes the final response in your brand voice — streamed directly into the Zendesk composer, ready to send or edit.
Two Modes, One Platform
Run RAUM as a fully automatic assistant that closes tickets in the background, or as a smart Copilot that drafts replies for your team to review — or both, side by side.
Copilot Mode
RAUM drafts replies inside the Zendesk composer. Your agents review, edit, and send — with full control over every response.
- AI Reply Generation — real-time streaming
- Grammar Fix & Translate — auto-detects customer language
- Ticket Enrichment — sentiment, category, confidence
- Custom Fields & Macros — read and write Zendesk fields
- Gallery — AI-curated image attachments for visual tickets
- Response Cache — instant replies for repeated queries
- Unlimited usage — no caps on any plan
Agent Mode
RAUM closes tickets end-to-end in the background, without a human. Your team only sees the ones that need their attention.
- Full pipeline execution — no human needed
- Tool Execution — trigger actions automatically across your systems
- Configurable overage — pause or continue
- Metered by ticket volume, not per agent
Intelligence & Customization
RAUM adapts to your organization — not the other way around.
Brand Voice Control
Configure tone, style, and response patterns per organization. The AI follows your exact instructions — not generic defaults.
Decision Trees & Knowledge Bases
Custom decision trees and knowledge bases that the PLANNER agent uses to route tickets through the right resolution path.
Zendesk Tag & Context Reading
Uses ticket metadata, tags, custom fields, and full conversation history for context-aware responses that understand the full picture.
Multi-Provider AI
Connect OpenAI, Anthropic, or Gemini. RAUM supports failover between providers so you're never dependent on a single model.
Analytics & Insights
Real-time analytics for your AI-powered support operation. Track costs, resolution rates, agent performance, and AI behavior.
Cost Per Ticket
Track avg cost per ticket with daily trends across all AI operations.
Mode Breakdown
See Agent Mode vs Copilot split with ticket counts, avg response time, and cost.
Token Usage
Breakdown by type — generate, translate, rewrite, refine — with cost attribution.
Activity Heatmap
7-day × 24-hour grid showing when your support operation is busiest.
CSAT Tracking
Satisfaction scores, good/bad ratings, resolution times, and reopening rates.
Knowledge Gaps
Auto-detected topics where AI had low confidence or no procedures matched.
Resolution Stats
P50/P95/P99 response time percentiles, counts by type and status.
Agent Usage
Per-user ticket distribution, token consumption, cost attribution, and last active.
Response Cache
Instant replies for repeated queries. Cache hits skip AI processing entirely, reducing cost and latency to near zero.
Integration
Works where your team already works. RAUM lives inside your Zendesk interface — no new tabs, no context switching.
Zendesk Native Experience
One-click install from the Zendesk Marketplace. No custom code required.
AI responses appear directly in the Zendesk ticket composer. Agents review, edit, and send.
Auto-Resolve handles tickets in the background, end-to-end. Your team only sees the ones that genuinely needed a human.
Secure API token authentication.
Simulation Mode
Test your entire AI pipeline against real tickets — without executing any real actions. Ship with confidence.
Dry-Run the Full Pipeline
Simulation Mode runs the full four-step pipeline — Understand, Plan, Act, Reply — against any Zendesk ticket. Lookups (knowledge base searches, data fetches) run for real so you get authentic context. Actions that would change state (order updates, ticket edits, external API calls) return a detailed preview of exactly what would have happened, without touching any real systems.
Real data, safe actions
Data lookups run live. Write operations are blocked and previewed as stubs.
Full agent trail
See every reasoning step, tool call, and decision the AI made — in one JSON response.
Test before you deploy
Validate new procedures, tool connections, or model changes against real tickets before going live.
Admin-only access
Simulation Mode is restricted to org admins, keeping your production environment safe.
Security & Compliance
Your customer data never leaves your control. Every layer of the pipeline is designed for security-first operations.
- Personal Data Protection
Names, emails, phone numbers, and other personal details are automatically masked before any data reaches the AI — and stay redacted in logs.
- AES-256 Encryption
All credentials and API keys are encrypted at rest using AES-256-GCM via AWS KMS, with automatic key rotation.
- Strict Data Separation
Your company's data is fully isolated from every other customer. Nothing is shared, mixed, or accessible across accounts.
- Full Audit Trail
Every AI action, admin change, and data access is logged with timestamps, IP addresses, and change history — ready for compliance reviews.