⚡ CLOSED-LOOP OPS INTELLIGENCE

The next on-call
is an agent.

Dera is the AI Operating System for engineering organizations. We provide an AI-native on-call platform that integrates with your incident management and communication systems to automatically resolve incidents, reduce MTTR, and build an Intelligent Memory of your infrastructure. Deployable as SaaS or Local Environment.

Start resolving for free → Book a demo
14-DAY FREE TRIAL • NO CREDIT CARD REQUIRED
INCIDENT RESPONSE debugging cloud-prod
# ecommerce-app-alerts
🚨
PagerDuty 8:26 AM
CloudWatch Alarm | RDS CPU utilization high | us-east-2
Threshold Crossed: 1 datapoint [99.6%] was greater than the threshold (80.0%).
🤖
Dera Agent 8:28 AM
✓ Investigation concluded

What happened: CPU spiked from ~9% to 99.6% at 8:27 AM. A single expensive SQL query saturated the db.r6g.large instance.

Customer impact: Recommendation service latency 25x (50ms → 1,244ms), throughput collapsed to ~20%.

Root cause: Feature flag database-health-monitor toggled ON at 8:25 AM, activating 15 CPU-intensive full-table scans per request.

Recommended action: Toggle flag OFF in etc/flagd/demo.flagd.json.

WHY DERA

Everything you need to debug manually
automate resolution.

01 - ON-CALL SERVICE

We take the page.

Dera participates in every on-call rotation to triage, investigate, and remediate alerts instantly. Backed by an MTTR SLA.

Alert
Agent
RCA & Fix
02 - CLOSED LOOP

End-to-end resolution.

Detects the issue, finds the root cause, generates a PR, monitors the deployment, and verifies the fix in production. If the issue persists, the loop continues.

Analyzing causal chain...
→ 08:25 AM: Deploy commit a4b9c1d
→ 08:26 AM: API latency spikes to 2s
→ 08:27 AM: OOM Kill on pod-x
✓ Root cause identified: Memory leak in worker pool
03 - INTENT-BASED ORCHESTRATION

Understands alerts dynamically.

Classifies incoming webhooks by domain and mode, routing to the best-matching agent with an appropriate budget and evidence policy.

Lookup
Investigate
Change
Explain
04 - INTELLIGENT MEMORY

Reason over your history.

Separates objective facts from subjective beliefs. Every incident, deploy, and PR review is structured into a memory architecture that agents use to adapt their behavior over time.

"We usually restart the ingestor pods when this queue backs up."
✓ Added to Agent Runbook
DEPLOYMENT

Two ways to deploy.
Zero setup work.

01 - SAAS

Hosted & Managed.

Connect your monitoring endpoints. Our cloud AI instantly starts debugging and resolving alerts via secure API integrations with transparent credential injection.

$ configure --endpoint=***
✓ Secured SSE Connection Active
02 - ON-PREM

Deployed in your infra.

Keep your data 100% secure. Deploy the intelligence engine entirely within your VPC. Perfect for highly regulated environments.

Your VPC
→
Local AI Engine
PRICING

Sell the service, not the software.

Pay As You Go

Per-Incident

Perfect for scaling teams. No baseline commitments, only pay for what Dera resolves.

  • ✓ Unlimited alert triage
  • ✓ Pay only for auto-remediations
  • ✓ Standard tool integrations
  • ✓ MTTR p95 ≤ 30 min SLA
  • ✓ Access to Intelligent Memory lite
Start 14-day free trial
START FOR FREE

Professional

$50/mo

First 2 months completely free. We replace a full level-1 on-call shift for your team with flat monthly predictability.

  • ✓ Unlimited alert investigations
  • ✓ Unlimited auto-remediations
  • ✓ Full Intelligent Memory ingest
  • ✓ Variance detection & Drift reporting
  • ✓ Multi-cluster agent management
Start 2 Months Free

Enterprise On-Prem

Custom

For large organizations with strict security, isolation, and compliance requirements.

  • ✓ Complete Local Deployment
  • ✓ Secure federation for internal APIs
  • ✓ BYOK & Zero data leakage
  • ✓ SOC2 Type 1 & HIPAA Compliance
Contact Sales
ABOUT US

Engineers should build.
Not babysit systems.

We built Dera because we were tired of being woken up at 3 AM just to query a database and restart a pod. By combining closed-loop intelligence with secure tool execution (MCP), we've created the first platform that doesn't just summarize alerts—it investigates, resolves them, and builds your Intelligent Memory.

The Dera Team
FAQ

Common Questions

What is Dera AI and how does it work? +
Dera AI is an AI-native Site Reliability Engineering (SRE) platform. It acts as an autonomous on-call agent that ingests alerts from tools like PagerDuty or Datadog, investigates root causes, and automatically generates pull requests or remediation steps to fix the incident, effectively reducing your Mean Time To Resolution (MTTR).
How does Dera AI differ from standard monitoring tools? +
Traditional monitoring tools only alert you when something goes wrong. Dera AI is a closed-loop resolution platform. It not only detects the variance but proactively investigates the underlying infrastructure, identifies the root cause, and applies a fix, replacing manual debugging.
How does the AI securely access our infrastructure? +
For SaaS, we use strict least-privilege API integrations. For maximum security, we offer our On-Prem deployment, running entirely inside your VPC with zero external network access required.
Will the AI accidentally break production? +
No. By default, the AI operates in "Read-Only RCA" mode, where it investigates and suggests fixes. You can explicitly authorize it to take actions (like restarting pods or creating PRs) by writing approval rules.
What environments do you support? +
We integrate out-of-the-box with cloud providers, container orchestration systems, monitoring platforms, incident management tools, and version control systems. You can easily add internal systems using standardized protocols.
Can the AI automatically rollback failed deployments? +
Yes. If authorized via your runbook rules, the agent can trigger rollback pipelines in your continuous integration and deployment pipelines when it detects critical regressions immediately following a deploy.
How does it handle false positive alerts? +
The agent doesn't just react to a single threshold. It cross-references the alert with broader system metrics, recent PRs, and database health to verify the impact before initiating any remediation or waking up an engineer.
What happens if the AI encounters an unknown edge case? +
If the agent cannot determine a high-confidence root cause, it halts execution and immediately escalates to the human on-call. It includes a summarized investigation report, saving you the first 15 minutes of manual log diving.

Ready to delegate your on-call?

Stop waking up for predictable alerts. Start automating your incident response today.

Book a Demo