Skip to content
DexCode

Parallel Editing

Edit multiple slides simultaneously using AI agent subprocesses for faster iteration.

What is Parallel Editing?

Parallel editing is DexCode's core differentiator. Instead of revising slides one at a time, you can assign edits to multiple slides simultaneously. Your AI agent spawns subagents — one per slide — that work concurrently.

How It Works

When you prompt an agent like Claude Code with a batch edit instruction, it:

  1. Identifies which slides need changes
  2. Spawns a subagent for each target slide
  3. Each subagent edits its assigned MDX file independently
  4. Hot reload updates the browser preview as each file is saved
bash
> Update slides 3, 5, and 8 in parallel:
  - Slide 3: Replace the bullet list with a comparison table
  - Slide 5: Add a pie chart showing market share data
  - Slide 8: Tighten the copy and improve visual hierarchy

Benefits

Traditional WorkflowParallel Editing
Edit slide 3, wait, reviewAll 3 slides edited simultaneously
Edit slide 5, wait, reviewReview all changes at once
Edit slide 8, wait, reviewSingle review cycle
~15 minutes~5 minutes

Best Practices

Be Specific

Give clear, actionable instructions for each slide:

bash
# Good — specific and actionable
> Slide 4: Replace "We're growing fast" with actual metrics.
  Use the revenue numbers from Q2: $1.2M ARR, 45% QoQ growth.
 
# Less effective — vague
> Make slide 4 better

Batch Related Changes

Group related edits into a single parallel request:

bash
> For slides 2-6, apply consistent styling:
  - Use the company's brand colors
  - Ensure all charts use the same color palette
  - Standardize heading sizes

Use AI Micro-Tuning

For final polish, delegate spacing, wording, chart labels, and visual details to AI with precision prompts:

bash
> On slide 7:
  - Align the chart axis labels to the grid
  - Reduce the padding between the title and subtitle
  - Make the footnote text 12px and 60% opacity

AI micro-tuning lets you ask AI to improve tone, hierarchy, spacing, and chart readability with precision prompts — achieving high-quality finishing touches without manual pixel-pushing.