Mohlo by se vám také líbit

Více informací o knize OBSERVABILITY

• Parametry knihy
• Anotace
• Oblíbené z jiného soudku

Nákupem získáte 38 bodů

Anotace knihy

What happens after you give an AI agent a goal?

The answer you see is only the surface. Between request and response, an agent may retrieve information, call tools, follow rules, use memory, retry failed steps, hand work to another agent, wait for approval, pass through guardrails, and take actions that affect the outside world. When that hidden journey is invisible, failures are harder to diagnose, improvements are harder to verify, and trust becomes guesswork.

OBSERVABILITY: Seeing What Your Agents Are Actually Doing is a first-principles guide to making agent behavior understandable.

Written for non-technical readers, the book starts with ordinary questions - What happened? Where did the work go? What changed? What evidence would prove it? - and only then introduces the technical language behind observability. Events become meaningful moments. Logs become useful records. Traces become connected journeys. Metrics become patterns that help you compare behavior across many tasks.

From that foundation, the book moves deeper into the realities of operating AI agents: instrumentation, correlation, spans, model and tool calls, latency, retries, baselines, failure patterns, evaluations, guardrails, incident response, audit trails, reliability objectives, rollouts, human review, memory and retrieval, permissions, budgets, multi-agent handoffs, drift, fleet-level observability, evidence quality, control testing, policy conformance, and earned autonomy.

The mathematics stays approachable. Rates, percentages, averages, medians, coverage, relative change, and simple reliability reasoning are used as practical tools for asking better questions - not as barriers to understanding.

Dialogues, fictional cases, visual frameworks, exercises, and a final Value Edition help turn the ideas into habits. You will practice separating evidence from explanation, breaking complex problems into smaller parts, testing rival hypotheses, designing an observability plan from a blank page, rehearsing incidents, and deciding whether the evidence is strong enough to support action.

This is not a vendor manual and not a coding textbook. It is a book about operational literacy for the agentic era: learning to see hidden work clearly enough to improve it, govern it, and know when greater autonomy has actually been earned.

If AI agents are beginning to do meaningful work in your organization - or you simply want to understand how responsible agent systems can be inspected and improved - this book gives you a practical language for seeing what is happening beneath the final answer.

Parametry knihy

• Plný název: OBSERVABILITY
• Podnázev: Seeing What Your Agents Are Actually Doing
• Autor: RAVINDRA KUMAR NAYAK
• Jazyk: Angličtina
• Vazba: Brožovaná
• Počet stran: 256
• EAN: 9798172265075
• ID: 53791945
• Nakladatelství: Independently published
• Hmotnost: 315 g
• Rozměry: 229 × 152 × 16 mm
• Datum vydání: 06. September 2026

---

---

---