AI Agents for Ops – Automated Incident Response (AIOPS)

AI - Artificial Intelligence, AI for Professions

In this practical hands-on course (30% theory, 70% practice) you will learn to apply multi-agent AI systems for automated incident investigation and automated resolution in IT operations. Sessions run in a prepared Linux lab, with emphasis on security and cost control.

This course explains agent architecture and the observe/decide/act loop, plus the subagent pattern to reduce blast radius and costs. You will run agent teams, design MCP safety boundaries, execute layered simulated incidents, plus production patterns for logging and escalation.

THIS TRAINING COURSE WILL HELP YOU:

  • Understand AI agents and how they differ from chat or scripts
  • Assemble an AI agent team for incident investigation
  • Use agents safely so they cannot perform unauthorized actions
  • Decide when an agent is appropriate vs script or monitoring

WHO SHOULD ATTEND?

  • Ops, DevOps, and SRE engineers wanting practical agent skills
  • Platform and infrastructure engineers evaluating agent deployment
  • Engineers working with infrastructure who want multi-agent Ops patterns
  • Not for developers focused on AI coding workflows
  • Not for data scientists or ML engineers — different discipline

COURSE LOCATION AND AVAILABLE DATES



Choose whether to attend in person in our classroom or join online. You can select your preferred format during registration. Learn more about hybrid training.

Unlock your employees' potential. We can adapt every course in our portfolio to your objectives and participants.
Need training at your premises or want to tailor the content and duration? We will prepare the right solution, in Czech or English.

Request customised training

Course content:

Hide details
  • Module 1 — Agent architecture for Ops
    1. Intro: what LLMs do, what chat can and cannot do
    2. What is an agent: observe / decide / act loop; difference from chat (tools + autonomy)
    3. When NOT to use an agent: scriptable tasks, deterministic processes
    4. Subagent pattern: why not one big agent — blast radius, costs, focused context
    5. Three roles: coordinator (strong model), syscheck + logcheck (cheaper models)
    6. Hands-on: build a coordinator and two specialist agents; run on a simulated alert
  • Module 2 — MCP servers as a safety boundary
    1. Problem from Module 1: agents with full shell access are unacceptable in production
    2. What is an MCP: networked tools, not shell access; like sudo rules — allowed commands only
    3. Two tool types: read-only (investigation) vs write (action)
    4. MCP anatomy: entry point, tool definitions, input/output schema
    5. Hands-on — Method 1: plug an existing syscheck-MCP into a syscheck profile
    6. Hands-on — Method 2: build a logcheck-MCP using a dev-squad (developer + tester + security subagents)
  • Module 3 — Automated investigation
    1. Investigation loop: trigger → investigate → report → (optionally) act
    2. When to stop, when to act, when to escalate
    3. Triggers: polling as a simple start (why start simple, not webhooks)
    4. Live demo: full chain on a prepared incident
    5. Hands-on: connect a trigger, run the full team on a layered real scenario (Disk full → MySQL lock → HTTP 500)
    6. Scenario design: layered checks require coordinator to correlate syscheck + logcheck outputs
  • Module 4 — Production patterns and cost control
    1. Model tiering: coordinator vs workers; real cost numbers for pipeline runs
    2. Failures and mitigations: infinite loops, token burn, hallucinated actions, cascading delegation, stale context
    3. Production checklist: read-only default, human-in-the-loop for write, cost budget, logging, alerting-on-alerting, graceful degradation
    4. When NOT to use agents: deterministic tasks, compliance-critical actions, tasks without an audit trail
Prerequisites:
Basic Linux server skills (SSH, shell, reading logs) and prior LLM chat experience.
Recommended previous course:
Linux – Basic Administration (LNX1)
Schedule:
1 day (9:00-17:00)
Price per person:
316.00 € ( 382.36 € incl. 21% VAT)

Training and learning environment