Pre-Read: Before the Alex Workshop
Read this the day before the workshop. It takes about 15 minutes. No note-taking required — just read and reflect.
What This Workshop Is About
You’re going to spend 90 minutes learning how to have a more productive conversation with an AI partner. Not prompts. Not automation. A conversation — the kind where you guide, refine, push back, and build something together.
This is different from what most people do with AI today. Most people issue commands and accept whatever comes back. This workshop teaches a better way.
From Prompt Engineering to Dialog Engineering
When AI tools first became popular, the skill everyone wanted was prompt engineering — the art of writing the perfect single instruction to get the AI to do what you wanted.
It works for simple tasks. “Summarize this paragraph.” “Translate this sentence.” “Give me 5 ideas for a title.”
But prompt engineering breaks down quickly when the task is complex — when you need nuance, context, iterative refinement, or judgment. A single well-crafted instruction cannot substitute for the ongoing back-and-forth of a real working relationship.
Dialog Engineering is the evolution. Instead of optimizing a single input, you treat the AI as a collaborative thinking partner. You set the context, gather information, structure the approach, generate content iteratively, and refine through feedback loops — the same way you would work with a talented colleague on a difficult problem.
The difference is fundamental:
| Prompt Engineering | Dialog Engineering |
|---|---|
| One input → one output | Many turns → refined output |
| Static instructions | Dynamic conversation |
| Works for simple tasks | Works for complex knowledge work |
| AI as a vending machine | AI as a thinking partner |
| Optimize the command | Build shared context over time |
The Research Behind This
This workshop is grounded in published research on AI-human collaboration. Studies consistently show that when professionals use AI in an iterative, context-aware way, they spend significantly less time on repetitive knowledge work — freeing capacity for more creative and strategic thinking.
The key word is iteratively. AI doesn’t automatically deliver this benefit. It delivers it to people who know how to work with it in an iterative, structured, context-aware way — the way this workshop teaches.
The research also identifies the primary failure modes: AI hallucinations (plausible but incorrect outputs), lack of contextual depth when given vague prompts, and outputs that require significant human correction when users don’t iterate. All of these are addressed by Dialog Engineering.
How a Dialog Engineering Session Works
A complete working session with an AI partner follows five phases — whether you’re writing a research paper, a technical specification, a campaign brief, or a strategic document.
Phase 1: Set the Scenario
Define the objective before asking for any output. Give the AI your role, the project context, and what specific outcome you need. This is the most important phase — it determines the quality of everything that follows.
The failure mode to avoid: Jumping straight to the task without context. (“Write me a report on X” vs. “I’m a research analyst preparing a briefing for a regulator on X. Here’s what I know…”)
Phase 2: Gather and Ground
Ask the AI to organize existing information, identify gaps, or structure what you’ve told it. This builds shared working context between you and the AI before any draft content is created.
Phase 3: Structure the Approach
Before generating the actual deliverable, agree on the structure. Ask the AI to propose an outline, framework, or plan — then react to it. “Add a section on Y.” “Remove Z — that’s out of scope.” This ensures the final output follows a shape you actually need.
Phase 4: Generate Iteratively
Tackle the deliverable section by section, turn by turn. Never accept the first output as final. Each turn adds precision. The conversation is the product.
Phase 5: Refine and Verify
Review the output as the intended reader — or ask the AI to do so from a specific perspective. Ask it to challenge its own reasoning. Ask it to identify what’s missing. The last turn is often as important as the first.
The Five Patterns You’ll Practice
In the workshop you will practice five specific conversation patterns. Here is a preview so the session feels familiar when you arrive:
1. Context-Goal-Constraints The foundation of every good AI request. Structure it as: “I’m a [role], working on [context]. I need [outcome]. Constraints: [limits].” This single change eliminates the most common source of weak AI output — insufficient context.
2. Explain-Like Tailor AI explanations to your actual knowledge level: “Explain this like I’m a [role] who knows [X] but not [Y].” Removes jargon and grounds the response in terms that are meaningful to you.
3. Show-Don’t-Tell Move from abstract to concrete: “Show me an example of [concept] applied to [my specific situation].” Forces the AI out of generic territory and into your actual context.
4. Iterate Treat every first output as a draft. “Good, but adjust [this]. Keep [that].” Build precision through successive turns — the conversation is the product, not the first response.
5. Challenge-Me Use the AI as a critical thinking partner: “What am I missing? What are the counterarguments?” This is where dialog engineering goes beyond what any single prompt can do.
What This Means for Your Work
Dialog Engineering is not an abstract productivity concept. It is a specific skill that applies to the kind of knowledge work you do every day — wherever that intersects with written analysis, structured documents, presentations, code, research, or communication.
You do not need to become an “AI person.” You need to become someone who can brief an AI the way you would brief a new colleague: with context, expectations, constraints, and feedback.
That is what this workshop teaches.
Before You Come
Spend 5 minutes thinking about one real task you’re currently working on — something you have to produce in the next two weeks. A report, a document, a presentation, an analysis, a specification. Anything substantive.
You’ll use it in the hands-on exercises. The more real it is, the more you’ll get out of the session.
One Question to Reflect On
In the last month, how often did you accept an AI output without pushing back on it — even when you knew it wasn’t quite right?
If the answer is “often,” that pattern ends today.
See you in the workshop.