liveBook · Manning

Failure mode

A failure mode is a distinct way in which a system, component, process, or operation can fail or produce an undesirable result.

How 6 Manning books explain it

The Art of AI Product Development

A failure mode is a distinct way an AI system can produce an incorrect, harmful, or otherwise undesired result.

How it is used
Frames failure modes as a starting point for managing AI errors in the user interface.
How deep it goes
Distinguishes error types and examines user perception and tolerance, without detailing implementation of error handling.
Examples
Illustrates errors with movie-review sentiment scores and an unrealistic brainstorming suggestion that sparks an idea.
What you can do after
Evaluate possible AI errors by type, user visibility, and tolerance in a given situation.

Read how The Art of AI Product Development explains failure mode →

Retrieval Augmented Generation, The Foundational Ideas

A failure mode is a specific way a system or component can behave incorrectly or fail to meet its requirements.

How it is used
Turns production RAG accuracy complaints into identifiable failure points for testing and targeted repair.
How deep it goes
Defines a three-layer lifecycle taxonomy and seven failure points, then links diagnosis to remedies rather than specifying a universal repair procedure.
Examples
Uses financial-services policy answers, missing or outdated guidelines, fabricated compliance procedures, and multi-hop and temporal questions.
What you can do after
Diagnose/debug with it by measuring the failing stage and selecting a proportionate remedy instead of adding complexity by default.

Read how Retrieval Augmented Generation, The Foundational Ideas explains failure mode →

AI Model Evaluation

A failure mode is a recurring way in which a model or system fails under particular conditions.

How it is used
Diagnoses where models, agents, or evaluators break across tasks, users, and operating conditions.
How deep it goes
Explains how to identify offline failure patterns, map them to live metrics, and set alerts; it does not prescribe a universal metric.
Examples
Uses support and travel-agent tasks, LLM-judge biases, and a Batman recommendation watchpoint to illustrate failures.
What you can do after
Apply offline findings by defining online watchpoints, alert thresholds, and pause-or-roll-back decisions.

Read how AI Model Evaluation explains failure mode →

Crack Any Codebase with AI

A failure mode is a specific way in which a software component or interaction can malfunction.

How it is used
Maps architectural components and connections to concrete ways they can malfunction.
How deep it goes
Classifies failure modes by component and connection type, but stops short of mitigation procedures or implementation internals.
Examples
Uses checkout-to-Stripe latency, queue-to-worker silent failure, database slow queries, cache stampedes, and frontend state errors.
What you can do after
Diagnose architectural risk by labeling services and wires, then asking whether users wait and whether connections cross the internet.

Read how Crack Any Codebase with AI explains failure mode →

LLM Customization and Fine-Tuning

A recurring way in which a retrieval-augmented generation system can produce incorrect or degraded behavior in production.

How it is used
Diagnoses production RAG problems by connecting business symptoms to specific system or model behaviors.
How deep it goes
Distinguishes four failure types and gives targeted checks and fixes for each.
Examples
Uses ignored retrieved context, wrong chunks, retrieval gaps, and rising latency as production cases.
What you can do after
Diagnose RAG failures by inspecting indexed chunks, testing context use, checking retrieval support, and monitoring latency.

Read how LLM Customization and Fine-Tuning explains failure mode →

Prompt Engineering in Practice

A failure mode is a distinct way in which a system, artifact, or process can produce an incorrect or undesired result.

How it is used
Diagnoses how prompt components affect outputs during prompt design.
How deep it goes
Explains partial failures through missing or weakened components, while deferring formal component vocabulary.
Examples
Uses incident reports to contrast generic summaries with engineering post-incident summaries.
What you can do after
Diagnose how unclear instructions, unlabeled inputs, or weak output guidance affect results.

Read how Prompt Engineering in Practice explains failure mode →

Also covered in

What to know first

Related concepts

Questions about failure mode

What is failure mode?

A failure mode is a distinct way in which a system, component, process, or operation can fail or produce an undesirable result.

Which Manning books explain failure mode?

The Art of AI Product Development, Retrieval Augmented Generation, The Foundational Ideas, AI Model Evaluation, Crack Any Codebase with AI, LLM Customization and Fine-Tuning and Prompt Engineering in Practice.

What should I know before learning failure mode?

Manning books that teach failure mode usually introduce conversational ai, latency, language model and AI agent first.

What kinds of failure mode are there?

Kinds of failure mode include argumentative inflation, defensive entrenchment, false dilemma, hidden regret, indistinct structures and overfitting.

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.