Usage note
This note is the fuller slide-copy companion to Practitioner Workshop Deck Outline - Paired Engineering with AI.
Use it when the practitioner workshop deck needs more complete on-slide language and presenter notes before any final slideware is produced.
Keep Workshop Pack - Paired Engineering with AI as the facilitator and exercise layer behind this deck.
This note is now treated as part of the accepted locked markdown baseline for the practitioner workshop package unless a substantive audience or content gap appears.
Slide 1. Title
On-slide copy
Paired Engineering with AI
A practical workshop for software delivery teams
Presenter note
Set the tone early: this is not a hype talk and not a tool demo. It is a working session about how to use AI well in real delivery work.
Slide 2. Why this workshop exists
On-slide copy
Tool access alone does not teach good AI usage.
- teams are already experimenting
- quality and learning can drift without shared practice
- this session is about safer and more useful habits
Presenter note
The point is to remove the implicit message of “use AI more.” The session is about using AI better, especially in ways that improve understanding, review, and workflow quality.
Slide 3. The middle path
On-slide copy
This is not a choice between avoiding AI and handing the work to the model.
The useful middle path is:
paired engineering with review, verification, and learning built in
- keep people thinking
- improve flow without outsourcing judgment
- avoid fluent but untrustworthy shortcuts
Presenter note
This slide gives the workshop its stance. It should feel simple enough to remember and strong enough to steer later exercises.
Slide 4. The paired-engineering loop
On-slide copy
Question -> generate or compare -> verify -> revise -> learn
- asking the model is only one step
- verification and revision are part of the work
- learning is an intended outcome
Presenter note
If participants remember only one loop from the session, it should probably be this one. It keeps the process from collapsing into “prompt -> paste.”
Slide 5. Learning mode versus delivery mode
On-slide copy
Not every task should use the same AI pattern.
Learning mode
- explanation-first
- slower, but capability-building
- better for unfamiliar or fragile work
Delivery mode
- bounded acceleration
- stronger verification
- better for familiar and reviewable work
Presenter note
This slide is one of the most important in the deck. It gives practitioners permission to slow down when the work is unfamiliar instead of feeling like every AI interaction should be optimized for speed.
Slide 6. What changes safe usage
On-slide copy
Good AI usage depends on more than job title.
Safe usage changes with:
- judgment and verification ability
- task familiarity
- task risk
- verification difficulty
Presenter note
Use plain language here. The deeper model is helpful, but the audience mainly needs to understand why copying somebody else’s AI habit is not automatically safe.
Slide 7. Verification is not one-size-fits-all
On-slide copy
Code, tests, requirements, architecture, and incident reasoning do not verify the same way.
- some outputs can be tested quickly
- some outputs sound good but are hard to evaluate
- low-observability work needs stronger caution
Presenter note
This is where the paired-engineering model becomes concrete. The key teaching move is that verification must match the artifact, not just the confidence of the person reading it.
Slide 8. Developer example
On-slide copy
AI helps most when the work is bounded and the feedback loop is real.
Good starter examples:
- bounded debugging
- code explanation
- localized unit test drafting
Use extra caution on:
- unfamiliar frameworks
- architecture-heavy changes
- production-critical logic
Presenter note
Make this feel practical, not generic. Developer audiences usually understand quickly when the examples are close to real repo work.
Slide 9. QA, SDET, architecture, and product examples
On-slide copy
Multiple roles can use AI, but not in the same way.
- QA or SDET: failure triage and test idea generation
- architects: option comparison and question generation
- product: backlog clarification and ambiguity detection
- every role has its own verification traps
Presenter note
This slide prevents the deck from sounding developer-only. It should also reinforce that different roles absorb different kinds of risk.
Slide 10. Common anti-patterns
On-slide copy
Most bad AI usage is easy to recognize once it has a name.
- asking for the full answer before understanding the problem
- trusting fluent output as proof
- using AI on hard-to-verify work without stronger review
- copying high-leverage habits without matching oversight ability
Presenter note
This is a good discussion slide. Participants usually recognize themselves or their teams in at least one of these patterns, which makes the rest of the workshop more real.
Slide 11. Exercise setup
On-slide copy
We learn good usage by practicing decisions, not just hearing advice.
In the exercises:
- decide mode first
- decide verification path next
- compare choices by role and task
Presenter note
This is the transition from concept to practice. Keep the instruction simple and let the workshop pack handle the specific scenarios and facilitator flow.
Slide 12. Debrief and team commitments
On-slide copy
Good enablement changes habits in real work.
Each participant should leave with:
- one workflow to pilot
- one anti-pattern to avoid
- one verification habit to strengthen
Presenter note
The workshop should end with concrete behavior change, not just agreement. This slide should make the session feel actionable instead of inspirational-only.
Slide 13. Keep practicing after the workshop
On-slide copy
The worksheet packs are the easiest next step after this session.
- start with the pack that matches your level
- use the size and complexity labels to choose a realistic scenario
- if you do not complete one yet, at least review one with a peer or lead
Presenter note
This is the slide that turns the exercise layer into a real call to action. The ask should feel reasonable. Review is enough to start. Completion is better when time allows.
Slide 14. Where to go deeper
On-slide copy
The workshop is the start, not the full delivery model.
Go deeper with:
- practitioner playbook
- verification standards
- capability model
- worksheet pack
- exercise library
- follow-up role-specific guidance
Presenter note
Point participants toward the durable notes. This helps the workshop function as an entry point into the broader project materials rather than a standalone event.
The new exercise library is especially useful when teams want a more deliberate progression for junior and intermediate practitioners instead of repeating the same starter scenario. The same library now also includes advanced senior and Staff scenarios focused on standards, tooling, rollout judgment, and low-observability reasoning.