25. Production Readiness Index¶

Why This Chapter?¶

You completed the main course and understand agent basics. Now you need to transition from a "learning agent" to a "production agent" that works reliably, safely, and efficiently in real environments.

This chapter serves as a prioritization guide and quick reference for production topics. It helps you understand what to implement first and provides quick links to detailed chapters.

Real-World Case Study¶

Situation: You built a working agent locally. Want to deploy it to production, but don't know where to start.

Problem:

Too many production topics to study at once
Don't know what's critical vs. nice-to-have
Need prioritization guide

Solution: This index provides a prioritized roadmap: start with mandatory blocks (observability, cost control, security), then add topics as you grow.

Prioritization Guide¶

Mode 1: Urgent to Production in 1 Day (Minimal Set)¶

If you need to launch an agent to production right now, start with these three topics:

19. Observability and Tracing — Without this you're blind. Needed immediately.
20. Cost & Latency Engineering — Critical for budget control.
17. Security and Governance — Mandatory production block.

These three topics give you basic production readiness: you'll see what's happening, control costs, and protect the system.

If you have time for planned refinement, add topics as you grow:

Week 1:

21. Workflow and State Management — When agents execute long tasks
22. Prompt and Program Management — When prompts change frequently
23. Evals in CI/CD — Automatic quality checking

Week 2:

24. Data and Privacy — If working with personal data

Production Topics Overview¶

Mandatory Production Blocks (Needed Immediately)¶

19. Observability and Tracing ¶

When needed: Immediately, as soon as agent goes to production.

What's inside: Structured logging, tracing agent runs and tool calls, metrics (latency, token usage, error rate), log correlation via run_id.

20. Cost & Latency Engineering ¶

When needed: When agent is used actively or works with large contexts.

What's inside: Token budgets, iteration limits, caching, fallback models, batching, timeouts.

17. Security and Governance ¶

When needed: Immediately, as soon as agent goes to production.

What's inside: Threat modeling for tool agents, risk scoring, prompt injection protection, RBAC for tools, dry-run modes, audit.

Topics as You Grow¶

21. Workflow and State Management in Production ¶

When needed: When agents execute long tasks (minutes or hours), need idempotency or error handling with retry.

What's inside: Queues and asynchrony, scaling, distributed state. Basic concepts (idempotency, retries, deadlines) described in Chapter 11: State Management.

22. Prompt and Program Management ¶

When needed: When prompts change frequently, there are multiple versions, or need A/B testing.

What's inside: Prompt versioning, prompt regressions via evals, configs and feature flags, A/B testing.

23. Evals in CI/CD ¶

When needed: When prompts or code change frequently and need automatic quality checking.

What's inside: Quality gates in CI/CD, dataset versioning, handling flaky cases, security tests.

Specialized Topics¶

24. Data and Privacy ¶

When needed: When agent works with personal data (PII) or secrets.

What's inside: PII detection and masking, secret protection, log redaction, log storage and TTL.

Note: Production aspects of RAG and Multi-Agent are described in basic chapters 06 and 07. Model selection and decoding configuration are described in Chapter 01 and Lab 00.

Prioritization Algorithm¶

Don't try to study everything at once. Use this algorithm:

Start with mandatory production blocks:
Observability (logging, tracing) — needed immediately, without this you're blind
Cost & latency engineering — critical if agent is used actively
Security and Governance — mandatory production block
Add topics as you grow:
Workflow/state — when agents execute long tasks or need idempotency
Prompt/program management — when prompts change frequently or there are multiple versions
Evals in CI/CD — when need automatic quality checking
Specialized topics:
Data/privacy — if working with personal data
RAG/Multi-Agent production aspects — see production notes in Chapter 06 and Chapter 07

Connection with Other Chapters¶

Basics: Basic concepts studied in Chapters 01-12
Advanced Patterns: Ecosystem and patterns studied in Chapters 13-18
Production Topics: Detailed production guides in Chapters 20-24

What's Next?¶

After implementing production readiness, your agent is ready for deployment in the real world. Continue monitoring, iterations, and improvements based on production metrics and user feedback.

Navigation: ← Data and Privacy | Table of Contents | Appendix →