Written by: Nimesh Chakravarthi, Co-founder & CTO, Struct
Key Takeaways
- Cleric AI delivers reactive SRE on-call triage in Slack with confidence-scored investigations, but it still needs manual prompts and setup that often exceeds 10 minutes.
- Setup follows 6 core steps: Slack and PagerDuty authentication, observability integrations, custom playbooks, testing, on-call schedules, and feedback loops.
- Cleric’s workflow covers alert-triggered investigations, auto-triage dashboards, conversational RCA, approved actions, and learning from resolutions, yet it remains read-only.
- Struct outperforms Cleric with proactive auto-investigations that cut investigation time from 45 minutes to 5, fast setup, auto-PR creation, and deeper automation across key metrics.
- Track MTTR reductions and avoid pitfalls like poor logs. Automate your on-call runbook with Struct for complete engineering team relief.
How Cleric AI Supports SRE On-Call Automation
Cleric AI launched the first self-learning AI SRE agent on December 9, 2025, designed to autonomously investigate incidents and deliver findings directly in Slack with confidence scores. The platform correlates context across logs, metrics, and configurations rather than reacting to individual alerts in isolation.
Early adopters like BlaBlaCar report reducing time lost to repetitive troubleshooting, which shows clear value for teams focused on investigation efficiency. These gains come with constraints, because Cleric operates as a reactive system that requires manual prompts and setup times exceeding 10 minutes. The platform also functions in read-only mode, which limits its ability to execute automated fixes and keeps remediation work in engineers’ hands even after analysis completes.
For teams demanding 80% faster, proactive results that reduce investigation time from 45 minutes to 5 and auto-investigate before laptop wake-up, see how Struct delivers investigations before you wake up.
How to Set Up Cleric AI for SRE On-Call (6 Steps)
Understanding Cleric’s setup requirements reveals why its reactive approach demands more configuration effort than proactive alternatives. Setting up Cleric AI requires connecting multiple observability and incident management tools.
1. Sign up and invite team members
Create a Cleric account and authenticate with Slack and PagerDuty for incident routing.
2. Connect observability platforms
Integrate read-only API keys for Datadog, Grafana, Prometheus, and cloud providers such as AWS, GCP, and Azure.
3. Configure custom playbooks
Enter specific correlation ID formats and on-call runbooks that guide Cleric’s investigation logic.
4. Test alert simulation
Run test incidents to validate confidence thresholds and confirm investigation accuracy.
5. Assign on-call schedules
Configure PagerDuty integration so alerts route automatically to the correct team members.
6. Monitor and provide feedback
Review investigation results regularly and train the system through structured feedback loops.
The following table summarizes the core integrations that support Cleric’s investigation workflow. Each integration supplies a different slice of context that the AI uses to assemble a complete incident picture.
| Tool | Integration Type |
|---|---|
| PagerDuty | Incident Management |
| Datadog | Observability |
| GitHub | Code Context |
| Slack | Communication |
Setup complexity and telemetry quality requirements can extend configuration beyond 10 minutes, particularly for teams with custom monitoring stacks.
Cleric AI in Action: Incident Triage and Self-Healing Workflow
Once configured, Cleric AI follows a reactive investigation workflow that activates when alerts fire.
1. Alert triggers investigation
Incoming alerts from PagerDuty or monitoring tools activate the AI agent in Slack channels.
2. Review auto-triage dashboard
Cleric presents ranked hypotheses with confidence scores and links to supporting evidence.
3. Query for root cause analysis
Engineers ask follow-up questions in Slack to explore additional context or test alternative theories.
4. Execute approved playbook actions
Human approval remains required for any remediation steps or system changes that affect production.
5. Document and handoff resolution
Investigation findings flow back into the learning system so future incidents benefit from past resolutions.
Cleric customers report significantly reducing investigation times through pattern recognition and faster context gathering. However, the reactive nature means engineers still wake up to manually guide investigations, and the system cannot automatically create pull requests or execute fixes.
Struct vs Cleric AI: Choosing Proactive SRE Automation
These workflow limitations, including manual prompting, read-only operations, and reactive triggering, directly affect the metrics that matter most for on-call teams. While Cleric provides reactive triage capabilities, Struct delivers proactive automation that completes investigations before engineers wake up. The following comparison highlights how Struct’s architectural approach turns into measurable advantages across the metrics that matter most for SRE teams.
| Feature | Struct | Cleric AI | Winner |
|---|---|---|---|
| Setup Time | 10 minutes | takes minutes | Struct |
| Investigation Speed | 80% reduction (45→5 min) | Manual prompting required | Struct |
| Approach | Proactive auto-investigation | Reactive triage | Struct |
| Integrations | Datadog, Sentry, AWS, GCP, Slack-native | Cleric supports integrations including Kubernetes, Datadog, Splunk, MongoDB Atlas, PagerDuty, AWS, GCP, Grafana, Prometheus, GitHub, Confluence, Elasticsearch, MotherDuck, and more. | Struct |
| Automation | Auto-PR creation, runbook execution | Read-only analysis | Struct |
Struct customers report 85-90% helpful investigation rates with proactive Slack notifications that contain complete root cause analysis before engineers open laptops. For teams requiring true automation rather than reactive assistance, explore Struct’s proactive approach.
Measuring SRE Automation Impact and Avoiding Pitfalls
Successful AI-driven SRE automation depends on tracking specific metrics and avoiding common implementation mistakes.
Key Metrics:
- Mean Time to Resolution (MTTR): Target reduction from 60-120 minutes to under 10 minutes so incidents resolve before they affect most users.
- Alert noise reduction: Measure the percentage of actionable alerts compared with false positives to understand signal quality.
- Engineering capacity recovery: Track hours reclaimed from manual triage work and reallocated to roadmap delivery.
Common Pitfalls:
- Poor log quality produces weak AI insights, so teams need structured logging and correlation IDs before expecting accurate root cause analysis.
- Over-reliance on automation without human oversight for complex incidents can cause misdiagnosis when the AI encounters new or rare failure patterns.
- Insufficient feedback loops prevent the system from learning which investigations were accurate, which keeps the same analytical mistakes appearing in future incidents.
Teams should establish pre-implementation baselines and review AI accuracy on a regular cadence. Struct mitigates these risks through intelligent handoff mechanisms and comprehensive onboarding support.
FAQ
How does Struct’s setup compare to Cleric AI?
Struct deploys in about 10 minutes with seamless Slack integration and automatic observability tool connections. Cleric requires longer setup times for multiple API integrations and manual playbook configuration across different platforms.
What compliance standards does Struct meet?
Struct maintains SOC 2 and HIPAA compliance, which meets security requirements for Seed to Series C companies handling sensitive data. The platform processes logs ephemerally and avoids persistent storage of proprietary information.
Can Struct work with VPC-locked environments?
Struct requires API access to observability tools and logs to perform effective root cause analysis. Organizations with strict on-premises requirements that block external log access should evaluate whether AI-driven automation aligns with their security constraints.
How quickly can teams onboard with Struct?
New engineering team members can begin handling on-call duties immediately because Struct’s automated investigation summaries provide comprehensive context. The system removes the tribal knowledge barrier that usually prevents junior engineers from responding effectively to incidents.
Does Struct support custom monitoring stacks?
Struct’s composable architecture lets teams encode specific runbooks and correlation patterns for their unique infrastructure. Custom widgets ensure that relevant data appears consistently across different alert types and services.
What ROI can teams expect from Struct?
Teams typically achieve the triage time improvements mentioned earlier, which frees senior engineers to focus on product development instead of firefighting. These productivity gains compound as automated investigations reduce alert fatigue and improve overall system reliability.
In 2026, AI-powered SRE automation has become essential for maintaining competitive product velocity while meeting strict SLA requirements. Cleric AI provides reactive triage capabilities, yet teams seeking proactive automation that works before engineers wake up often find Struct’s approach delivers stronger results. Schedule a demo to see how Struct gives your engineering team their nights back.
