Why We Built a DSL for Presentations (and Why You Should Care)
Most AI slide tools go straight from text to pixels. We took a different path: text → Brain (DSL) → pixels. Heres why the intermediate representation matters.
Most AI presentation tools follow a simple pipeline: take text in, push pixels out. This works for demos but breaks down at scale. Here is why we took a different approach.
The fragility of text-to-pixel
When an LLM generates slides directly, every regeneration is a coin flip. Change one paragraph and the entire deck layout might shift. There is no way to selectively update a single slide without regenerating everything.
Our approach: text → Brain → pixels
The Brain is a JSON-based intermediate representation. It captures the semantic structure of your content — learning objectives, concept blocks, assessments — without committing to a visual layout.
Why this matters
- Selective re-rendering: change one block in the Brain, and only that slide updates
- Multi-format output: the same Brain renders to PPTX, PDF, facilitator guide, or quiz bank
- Validation before rendering: structural and brand rules are enforced on the JSON, not the pixels
- Version control: Brain JSON diffs cleanly, unlike binary PPTX files
Learn more
Read the Brain DSL reference or explore the API documentation to see how the compiler pipeline works end-to-end.