AI Assisted Software Engineering

The companion site · 18 chapters · 13 verdicts · 159 key concepts

AI Assisted Software Engineering

Build Better Software with Human Creativity and AI

Coding agents now write much of the code. Someone still has to decide what to build, estimate it, manage its risks, review it, control its changes and know whether it works. This textbook teaches each classical method from its foundation, then tests it with an assumption audit: what still holds, what to recalibrate, what breaks and what inverts.

Explore the mapBrowse the chaptersLearn the audit
Cover of AI Assisted Software Engineering by Dr. Jerry A. Smith

One method for every chapter

The assumption audit

  1. 1Method
  2. 2Assumption
  3. 3Evidence
  4. 4Verdict
  5. 5Replacement
  6. 6Prediction

Then one more question: what would have to change about agents for this conclusion to stop being true? How the audit works →

MapSee the whole book on one wheelEighteen chapters colored by verdict, the 159 concepts they define, and the 77 places where chapters lean on each other. ChaptersWhat each chapter asks of youThe assumption each method rests on, the verdict, the objectives, the lab and the prediction it tests, and the key terms. The auditA method that outlasts the toolsThe six steps, the negation test, the decision rule for the four verdicts, the evidence ladder and a blank audit form. ToolkitTemplates you will reuse all termThe work log, the handoff package, the two-part risk register, the estimation worksheet, the evidence package and the claim-to-evidence ledger. ProjectManage a team of agents for a semesterMilestones M0 to M6, the handoff to a classmate, the ten predictions your data can test, and a 14-week schedule. InstructorsTeach the courseHow the book fits a standard catalog description, chapter dependencies, and how to request the Instructor's Guide.

Thirteen audits, four verdicts

Two hold, five recalibrate, four break, two invert

2 chapters

Holds

The assumption still describes agent work. The method stands as taught.

8. Coordination at Scale10. Design for Change, and for Agents

5 chapters

Recalibrate

The method survives with new parameters, rates, thresholds or a redefined quantity.

3. Life-Cycle Models When Iteration Is Cheap7. Project Management and Where Value Is Earned9. Risk Management12. Reviews13. Testing, Evaluation, and Tests as Steering

4 chapters

Breaks

The method needs new variables, records or steps. Changing its numbers is not enough.

5. Size Estimation6. Cost and Schedule Estimation After COCOMO14. Configuration Management When Agents Commit15. Reliability of Nondeterministic Systems

2 chapters

Inverts

The activity's purpose was set by a scarcity that has reversed. What was cheap is now scarce.

4. Requirements as the Primary Artifact11. Construction When Code Is Cheap

The count summarizes the judgments; it measures nothing. Each verdict is dated, names the assumption it tested, and ends with a prediction your own project data can check.

The book

Dr. Jerry A. Smith

AI Assisted Software Engineering: Build Better Software with Human Creativity and AI, by Dr. Jerry A. Smith. 602 pages in four parts: 18 chapters and eight appendices, with a running example (Dispatch, a small dispatch dashboard) and a semester project in which each student manages a team of agents and inherits a classmate's project halfway through.

Editions and contents About the author