Workshop: Building Robust, Maintainable Software Using Generative AI

This was a day-long workshop at SDD 2026, 15th May 2026.

Even Kent Beck is vibe coding these days. Everywhere you go, you’ll see software professionals getting excited about what they’re able to build using tools such as CoPilot, Cursor and Claude Code. But for every person getting excited and making claims such as a 10x increase in productivity, you’ll see someone else telling you how unreliable the results can be, and casting doubt on the claims of the cheerleaders. How can you create robust maintainable systems, if you didn’t write the code yourself and you don’t understand how and why it works?

The news is full of tech companies who mandate the use of AI to write code. Junior developers are becoming more and more reliant on AI, and recruiters are struggling to tell the difference between applicants that did or didn’t use AI to enhance their submissions. But the range of genAI options is bewildering and ever-changing. And what about the ethical impact, not only on the state of our industry but also on global warming?

Many seasoned developers can find themselves reluctant to embrace AI. We enjoy writing code! Why would we want an LLM to do it on our behalf? But it feels like it’s going to get increasingly difficult to get and hold a position as a software developer without using AI.

It’s time to take a deep breath and leap in. Find out how to keep the fun of solving problems with code, while maximising the usefulness of AI. Learn how to use test driven development and robust test suites to monitor genAI outputs, understand what they’ve created for you and maintain confidence in the quality of your software. Find out how to use small language models to minimise the environmental impact.

The workshop will involve hands-on coding as we build a simple app piece by piece, checking the results as we go and often starting again from scratch to investigate different approaches and discover what impact they have on the quality of the output. By combining TDD (Test Driven Development) with the outputs of an AI coding assistant, what you can get is something very powerful.

Topics covered will include:

Intro to genAI coding and current arguments for and against
Getting started: First choices and simple results
Small Language Models (SLMs)
Prompt engineering: how to persuade genAI to behave itself
Getting started with the test-driven approach
Comparing the test-driven approach with allowing GenAI to roam free
Quality assurance and risk mitigation
Tweaking our app for best results
Summary of what we’ve learned
Laptops required. Participants will work both independently and in pairs. Programming languages used will be the choice of the participants. You don’t need to be a coding expert, but we do assume you have the ability to build simple software systems in the language of your choice. All attendees will need an OpenAI subscription in order to participate fully in the workshop.

Transcript

1. Building Robust, Maintainable Software Using Generative AI @ClareSudbery
2. @ClareSudbery ALL LINKS FROM TODAY (including a link to these slides)
3. 9:30 – 10:30 @ClareSudbery
4. @ClareSudbery YOU MIGHT NOT STAY IN YOUR SEAT Be prepared to move to a different seat!
5. @ClareSudbery RAISED HAND MEANS QUIET PLEASE, ATTENTION NEEDED! 00:00 – 05:00 MANY
6. DEMO Q&A EXERCISE, PART ONE (30 mins) BREAK Agenda 9:30 – 10:30 10:30 – 11:00 11:00 – 12:30 (60 mins) LUNCH 12:30 – 13:30 @ClareSudbery DISCUSSION EXERCISE CONT. WRAP-UP EXERCISE, PART TWO (30 mins) BREAK 13:30 – 15:00 15:00 – 15:30 15:30 – 17:00 ENDS 17:00
7. How many people? @ClareSudbery
8. @ClareSudbery Windsurf – free until you reach your token limit (which could happen during the day!) Cursor – free up to a limit, or if you have your own API keys from OpenAI or Anthropic (Claude) VSCode with GitHub CoPilot – free up to a limit Download a tool (if you haven’t already) One that will work in an integrated IDE
9. Intro @ClareSudbery
10. @ClareSudbery 2 minutes Write your most optimistic LLM- coding goal on a sticky note What would be fun to build using LLM-augmented coding? Write clearly – people need to read it Add your name and remember what you wrote!
11. @ClareSudbery This half of the room, put your sticky notes up Spread them out along the wall
12. @ClareSudbery YOU ARE LIKELY TO BE ABOUT TO MOVE SEATS Be prepared! And remember to introduce yourself… Name + role
13. @ClareSudbery Other half of the room, take one sticky note each One that you like the look of
14. @ClareSudbery Walk around, waving your sticky note Find the person who wrote it Form a pair / three with them Introduce yourselves! Name + role
15. TDD • Know what it is? • Used it before? • Use it regularly? @ClareSudbery
16. @ClareSudbery Test Driven Development
17. bit.ly/making-tech-better • Geepaw Hill – Test Driven Development I hosted a podcast… @ClareSudbery
18. Exercise, Part One @ClareSudbery
19. All the resources we use today are available in the following GitHub repo: github.com/claresudbery/ai-tdd-workshop • Exercise descriptions (exercises folder) • Sample markdown (markdown folder) • Sample results (results folder) • Sample process files (process-files folder) @ClareSudbery
20. USE SOURCE CONTROL github.com/claresudbery/ai-tdd-workshop • So… Either FORK my repo • …or create your own repo • (there’s no working code in mine • - just reference files • - so you can just copy files from there) @ClareSudbery
21. @ClareSudbery Windsurf – free until you reach your token limit (which could happen during the day!) Cursor – free up to a limit, or if you have your own API keys from OpenAI or Anthropic (Claude) VSCode with GitHub CoPilot – free up to a limit Download a tool (if you haven’t already) One that will work in an integrated IDE
22. @ClareSudbery Open the tool you’re going to be using Ask your AI to help you build the following tool: • Take a markdown file as input – see markdown folder in repo for a sample (sample-input.md) • The markdown file is text from a Slack channel. • There are also 28 sample reply threads, in sub-folder in markdown folder • You want the tool to identify reply thread insertion points, and insert each reply thread in the correct place • You want a nice UI that allows users to upload either all the relevant markdown files, or paste text in directly • (WARNING: Text copied from Slack will be ugly. The sample files have had some processing.) • The output should be a nicely formatted markdown file. • DON’T ATTEMPT TO WRITE CODE YOURSELF. NO TESTS UNLESS LLM SAYS SO! LLM IS IN CHARGE. • These instructions are available in the GitHub repo, in exercises/part1-basic.md. • If you finish this exercise and want more, see exercises/part1-extensions.md. • Go!
23. Demo : An example of what we’ll try to make @ClareSudbery
24. Halfway (ish) Start a new chat context …because in the real world, projects are shared and contexts are lost @ClareSudbery
25. 10:30 – 11:00 BREAK @ClareSudbery
26. RESULTS • Who completed part 1? • Who completed the extension? • What problems did you encounter? • Who's happy with the result? What did you like? • If you were successful, why do you think that is? @ClareSudbery
27. CODING WITHOUT TESTS • Does your solution work? • How do you know it works? • How will you know if it's broken? • What if you start a new session and continue? • What if you want to add new features? @ClareSudbery
28. 11:00 – 12:30 @ClareSudbery
29. So, how CAN you combine AI with test driven development? @ClareSudbery
30. @ClareSudbery
31. DEMO @ClareSudbery
32. @ClareSudbery LLM-AUGMENTED CODING: GOLDEN RULES 1. Encourage the LLM to ask clarifying questions 2. Move in small steps. Big steps feel fast but they quickly slow you down. 3. Always start with tests. - a. Ask the LLM to write tests before code. - b. Check and refine the tests. - c. Ask the LLM to make the tests pass by implementing the solution. - d. Keep running all tests to make sure nothing is broken. 4. Ask the LLM to summarise conversations. Persist those summaries in files. 5. Use a starter character in each process file. 6. Throw things away! Start again from scratch. Beware the sunk cost fallacy. 7. Regularly start a new chat context. 8. Be polite. Use "please" and "thank you".
33. @ClareSudbery LLM-AUGMENTED CODING: GOLDEN RULES 1. Don't write code (first draft) if AI can do it for you 2. Consider a small language model 3. Ask WHY 4. Beware confirmation bias (in both directions) 5. Explore ways of measuring effectiveness
34. @ClareSudbery
35. 12:30 – 13:30 LUNCH @ClareSudbery
36. 13:30 – 15:00 @ClareSudbery
37. Exercise, Part Two @ClareSudbery
38. @ClareSudbery EXERCISE TWO STEP 1 • Get your LLM to help you summarise everything you’ve discussed so far, about what you’re building and how you want it to function • Persist the results – eg in a file called project_spec.md
39. @ClareSudbery EXERCISE TWO STEP 2 • Start again from scratch! • Delete everything except project_spec.md (or equivalent) • Start a brand new chat window. Try a different LLM if you can! • Tell the LLM you want to proceed using TDD (test driven development) • >>>
40. @ClareSudbery EXERCISE TWO STEP 2 (continued) • Ask it to read and follow instructions in testing_process.md and running-scripts.md or any other process files from the process- files folder in the GitHub repo. • Remember the golden rules. • These instructions are available in the GitHub repo, in exercises/part2.md.
41. 15:00 – 15:30 BREAK @ClareSudbery
42. 15:30 – 17:00 @ClareSudbery
43. Demo – tests and project spec for reply threads @ClareSudbery
44. Discussion Questions? Useful tips? 1-2-all @ClareSudbery
45. Exercise part 2, continued @ClareSudbery
46. Wrap-up – Did the LLM meet, exceed or disappoint your expectations? @ClareSudbery
47. @ClareSudbery CONCLUSIONS • Golden rules apply to people as well as LLMs! • Gotchas (exercise caution – I ain’t no cheerleader): • Confirmation bias • Context impacts code • Security • Out of date technologies • Data protection • Ethics • How much power the LLM has (to commit, edit, etc) • Environmental impact (small language models)
48. @ClareSudbery 2 minutes Write new stickies for either “INSPIRED” or “STILL SEEKING ANSWERS” Make sure you include the headings on the stickies! It will help me. Then stick them on the wall in the designated areas.
50. Landing in a neat pair-synchronized way. https://www.instagram.com/p/BsY1qBeg5 Qg @ClareSudbery
51. @ClareSudbery THANK YOU! FEEDBACK
52. @ClareSudbery ALL LINKS FROM TODAY (including a link to these slides)