AGENCYSCRIPT
CoursesEnterpriseBlog
đź‘‘FoundersSign inJoin Waitlist
AGENCYSCRIPT

Governed Certification Framework

The operating system for AI-enabled agency building. Certify judgment under constraint. Standards over scale. Governance over shortcuts.

Stay informed

Governance updates, certification insights, and industry standards.

Products

  • Platform
  • Certification
  • Launch Program
  • Vault
  • The Book

Certification

  • Foundation (AS-F)
  • Operator (AS-O)
  • Architect (AS-A)
  • Principal (AS-P)

Resources

  • Blog
  • Verify Credential
  • Enterprise
  • Partners
  • Pricing

Company

  • About
  • Contact
  • Careers
  • Press
© 2026 Agency Script, Inc.·
Privacy PolicyTerms of ServiceCertification AgreementSecurity

Standards over scale. Judgment over volume. Governance over shortcuts.

On This Page

The Model in BriefThree Stages, Three OwnersWhy Separation HelpsStage One: DraftWhat This Stage OwnsThe Input DisciplineStage Two: ShapeImpose the ArgumentEstablish TruthStage Three: RefinePolish Voice and DensityBrand and RehearseWhen to Apply the ModelScale It to the StakesThe Handoffs Are the DisciplineCommon Ways the Model Breaks DownCollapsing Draft and ShapeRefining Before ShapingSkipping the Read-AloudApplying the Model With a TeamAssign Stages to PeopleUse It as Shared LanguageFrequently Asked QuestionsWhy not just edit as I go instead of using stages?What distinguishes Shape from Refine?Can I skip Draft and outline by hand instead?How rigid is the sequence?Does the model change for different tools?How do I know when a stage is finished?Key Takeaways
Home/Blog/The Draft-Shape-Refine Model for Generated Slides
General

The Draft-Shape-Refine Model for Generated Slides

A

Agency Script Editorial

Editorial Team

·December 4, 2018·8 min read
AI presentation toolsAI presentation tools frameworkAI presentation tools guideai tools

Most people use slide generators as a one-shot machine: type, generate, present. That works for throwaway decks and fails for anything that matters. The fix is to stop treating generation as the whole job and start treating it as one stage in a repeatable model. This article introduces that model, Draft-Shape-Refine, and explains what each stage owns, how work passes between them, and when the model earns its overhead.

A model is valuable when it tells you what to do at each moment and, just as importantly, what not to do yet. Draft-Shape-Refine does this by assigning each stage a single responsibility. When you know which stage you are in, you know which decisions are in scope and which to defer. That clarity is what prevents the common failure of editing copy before the structure is even settled.

The three stages are sequential but not rigid; you will sometimes loop back. What matters is that each stage has a job, and you do not skip ahead to the next until the current one is done.

The name is deliberately plain because the value is in the discipline, not the label. You could call the stages anything; what matters is that you stop treating slide generation as a single undifferentiated act and start seeing it as a pipeline with distinct responsibilities. Most of the frustration people report with these tools, generic decks, fabricated facts, hours lost to rework, traces back to collapsing the pipeline into one step. The model's whole purpose is to keep the steps apart so each one does its job cleanly.

The Model in Brief

Three Stages, Three Owners

Draft is where the tool produces a complete first version from your input. Shape is where you, the human, impose structure and truth. Refine is where you polish voice, design, and delivery. Generation owns Draft; you own Shape and Refine entirely.

Why Separation Helps

The stages fail when they blur. Editing wording during Draft wastes effort on content you may cut. Fixing structure during Refine means redoing polish. Keeping the stages distinct means each piece of work happens once, at the right time.

Stage One: Draft

What This Stage Owns

Draft owns volume and structure. You feed the tool a goal, an audience, and an outline, and it returns a full deck. The output is raw material, not a product. Its job is to give you something to react to, fast.

The Input Discipline

The quality of Draft is set by your input, not the tool. A strong outline, led by goal and audience, produces a draft worth shaping. The mechanics of building that input are detailed in Turning a Rough Outline Into Finished Slides, Step by Step. Generate once; the goal is a starting point, not perfection.

Stage Two: Shape

Impose the Argument

Shape is where you take control. Override the tool's structure to serve your actual case, reorder slides, cut sections that do not advance the goal, and add ones the tool missed. This is the stage the one-shot users skip, and it is why their decks meander.

Establish Truth

Shape is also where the truth pass happens. Every number, name, and claim gets verified or deleted. This is non-negotiable, and the failures it prevents are catalogued in Where Generated Decks Go Sideways, and What Fixes Them. Leave Shape only when the deck's structure and facts are sound.

Stage Three: Refine

Polish Voice and Density

Refine is the editorial pass. Cut copy to one point per slide, rewrite anything that does not sound like you, and tighten transitions. By now the structure is fixed, so this work is safe, you are polishing content you have decided to keep.

Brand and Rehearse

Refine ends with the brand pass, applying your colors, fonts, and visuals, and a read-aloud rehearsal against the goal. The opinionated habits that make this stage consistent are in Habits That Separate Polished AI Decks From Sloppy Ones.

When to Apply the Model

Scale It to the Stakes

For a high-stakes client deck, run all three stages fully. For a throwaway internal update, you might collapse Shape and Refine into a single quick pass. The model flexes; the sequence does not. Never present from Draft alone.

The Handoffs Are the Discipline

The value is in respecting the handoffs: do not start Shape until you have a full Draft to react to, and do not start Refine until structure and truth are settled. A real deck running through this model is told in One Studio Rebuilt Its Pitch Deck With Generated Slides.

Common Ways the Model Breaks Down

Collapsing Draft and Shape

The most frequent failure is treating the generated Draft as if Shape were already done, accepting its structure and facts wholesale. This is the one-shot habit in disguise. The symptom is a deck that flows nicely to a generic conclusion. The cure is to consciously enter Shape, override the structure, and run the truth pass, even when the Draft looks finished.

Refining Before Shaping

The opposite failure is polishing wording and styling before the structure is settled. You spend effort perfecting slides you will later cut or reorder. The model exists partly to prevent exactly this: Refine work is only safe once Shape has locked the structure and the facts.

Skipping the Read-Aloud

Refine is not complete until you have heard the deck. Teams that end at the brand pass ship decks with unsayable lines and broken transitions. The rehearsal is the exit condition for Refine, not an optional extra, and skipping it surfaces problems live instead of in private.

Applying the Model With a Team

Assign Stages to People

The model maps cleanly onto collaboration. One person can own the outline and Draft, another can run Shape, and a third can handle the Refine polish and brand pass. Because each stage has a defined responsibility and exit condition, handoffs between people are clean rather than chaotic.

Use It as Shared Language

When a team agrees on the stages, feedback gets sharper. Saying a deck is still in Shape tells everyone not to comment on wording yet. Saying it is in Refine signals the structure is locked. That shared vocabulary prevents the circular reviews where people argue about polish while the foundation is still moving. A team that has run a real deck through this model is described in One Studio Rebuilt Its Pitch Deck With Generated Slides.

Frequently Asked Questions

Why not just edit as I go instead of using stages?

Because editing as you go means polishing content you may later cut and fixing structure after you have already styled it. The stages prevent rework by ensuring each kind of decision happens once, at the point where it is cheapest to make and hardest to undo.

What distinguishes Shape from Refine?

Shape is about structure and truth, what the deck says and in what order, while Refine is about polish, how it says it. Shape decisions are expensive to reverse, so they come first. Refine decisions are cosmetic and safe once the foundation is set.

Can I skip Draft and outline by hand instead?

You can, and for very short decks it may be faster. But Draft's value is producing volume to react to quickly. For most decks, generating a full draft from a strong outline beats writing every slide by hand, then shaping what you got.

How rigid is the sequence?

The order is firm but not a one-way street. You will occasionally return to Shape if Refine exposes a structural gap. What you avoid is starting Refine before Shape is done, since polishing a structure you will change wastes the effort.

Does the model change for different tools?

No. The model is about the division of labor between human and tool, which is constant. Whatever generator you use, it owns Draft, and you own Shape and Refine. The tool's features affect how Draft and the brand pass feel, not the stages.

How do I know when a stage is finished?

Draft is done when you have a complete deck to react to. Shape is done when structure and facts are sound. Refine is done when the deck reads cleanly aloud and serves its goal. Each stage has a clear exit condition, which is what lets you move forward without second-guessing.

Key Takeaways

  • Draft-Shape-Refine splits deck work into three stages: the tool drafts, you shape, you refine.
  • Keeping the stages distinct prevents rework, since each kind of decision happens once at the right time.
  • Draft owns volume and structure; its quality is set by your input, not the tool.
  • Shape owns argument and truth, the verification pass that one-shot users skip and pay for later.
  • Scale the model to the stakes, but never present from a raw Draft alone.

Search Articles

Categories

OperationsSalesDeliveryGovernance

Popular Tags

prompt engineeringai fundamentalsai toolsthe difference between AIMLagency operationsagency growthenterprise sales

Share Article

A

Agency Script Editorial

Editorial Team

The Agency Script editorial team delivers operational insights on AI delivery, certification, and governance for modern agency operators.

Related Articles

General

Prompt Quality Decides Whether AI Earns Its Keep

Prompt quality is the single biggest variable in whether AI delivers real work or expensive noise. The model matters, the platform matters — but the prompt you write determines whether you get a first

A
Agency Script Editorial
June 1, 2026·10 min read
General

Counting the Real Cost of Every Token You Send

Tokens and context windows sit at the intersection of AI capability and operational cost—yet most business cases treat them as technical footnotes. That's a mistake that costs real money. Every time y

A
Agency Script Editorial
June 1, 2026·10 min read
General

Rolling Out AI Hallucinations Across a Team

Most teams discover AI hallucinations the hard way — a confident-sounding wrong answer makes it into a client deliverable, a legal brief, or a published report. The damage isn't just to the output; it

A
Agency Script Editorial
June 1, 2026·11 min read

Ready to certify your AI capability?

Join the professionals building governed, repeatable AI delivery systems.

Explore Certification