Cover of "Navigating the AI World: Using and Building AI at Work" by Prasad Kukkala

Book

Navigating the AI World

Using and Building AI at Work

By Prasad Kukkala

Publication pendingFirst editionUpdated 8 September 2026

Read onlineDownload PDF

This book connects everyday AI use with the systems that make it possible, from context and company knowledge to tools, agents and physical infrastructure. It is written for working professionals who want to understand what sits underneath the tools they use, decide what to learn next, and recognize the risks before they paste, connect or automate.

The central recommendation is simple: start with a bounded work problem, establish the data and authority it requires, and add complexity only when evidence justifies it.

Who this book is for

Working professionals who use AI tools at work and want to understand what sits underneath them, decide what to learn next, and recognise the risks before they paste, connect or automate. Part A is for everyone; later parts are for people who build, evaluate or operate AI systems.

What you will learn

  • How language models actually answer, and what that means for trust
  • Where your data goes when you paste, connect or automate, and the five questions to ask first
  • How to let a bounded company problem drive the design of an AI system
  • Retrieval with permission, narrow tool authority and the MCP boundary
  • Evaluation, evidence for release, and designing for hostile input
  • What production AI services cost, how they fail, and how to operate them
  • A roadmap you can revise, with hands-on labs

Contents

Contents (28 chapters)

  1. Front matter
  2. How to use this book
  3. Read this first
  4. Part A · For everyone who uses AI at work
  5. 1Understand what you are using
  6. 2What is actually happening when it answers
  7. 3Know where your data goes
  8. 4Judgement, overreliance and the fluency trap
  9. Part B · Deciding what to learn
  10. 5Where to start when everything is moving
  11. 6Choose what to know, use and master
  12. 7Build a roadmap you can revise
  13. 8Put the technology in its place
  14. 9What it costs and why
  15. Part C · Building AI into work
  16. 10Let the company problem drive the design
  17. 11Retrieve evidence with permission
  18. 12Test confidential retrieval end to end
  19. 13Give tools narrow authority
  20. 14Understand the MCP boundary
  21. 15Control an agent outside the model
  22. 16Walk through an attempted overreach
  23. 17Evaluate the task and its failure modes
  24. 18Decide what evidence permits release
  25. 19Design for hostile and unreliable input
  26. 20Give risk decisions an owner
  27. 21Assemble the production service
  28. 22Operate for failures and change
  29. Part D · Models, platforms and hardware
  30. 23Separate training from inference
  31. 24Choose model adaptation and deployment
  32. 25Read the seven-layer map correctly
  33. 26Understand GPU and memory constraints
  34. Practice labs
  35. 27Practise with evidence
  36. 28Practise actions and failure recovery
  37. Back matter
  38. Glossary
  39. References
  40. Technical references and representative tools

Corrections and updates

  • 2026-09-08: Web edition corrected — fixed a spacing typo in Chapter 19. The downloadable First Edition PDF retains the original printed text.

Written by Prasad Kukkala, technology practitioner and technical writer.