Generate documents without regenerating their design
Designer-built master templates in Slides, PowerPoint, and Word. Data binds into placeholders. Layout, typography, and brand are preserved on every generation.
The problem with most document automation isn’t the automation. It’s that the design quietly degrades every time the document gets rebuilt. A line wraps differently, a margin shifts, a brand color drifts toward the default, and three iterations later the report no longer matches the brand book.
SourceToDocs treats the template as a fixed, designer-owned asset. Variable content binds into the template at generation time — never the other way around. Your designer’s master file is the source of truth for everything that should stay the same. The data is the source of truth for everything that should change. The boundary between them is enforced.
What it does
The template engine accepts a designer-authored master file as input. Today that means a Google Slides deck, a PowerPoint file (.pptx), or a Word document (.docx). Inside the master, your designer marks the places where data should land using a simple placeholder convention:
{{field_name}}for plain text that should be replaced — titles, dates, names, descriptions.[[BLOCK_NAME]]for image or shape placeholders that get replaced with media or a formatted block of content.- Section markers for conditional regions (show only if a sponsor exists; hide if there are no risks to report) and repeating regions (one master slide produces N output slides from a list).
At generation time, the engine walks the template, resolves every placeholder against the bound data, and writes the result to the requested output format. Layout, typography, brand palette, master slides, themes, headers and footers — none of it is regenerated. The structural elements your designer made stay exactly as they were.
This is the difference between SourceToDocs and prompt-to-deck AI tools. Tools like Gamma or Beautiful.ai generate the design alongside the content, which is useful for first drafts and one-offs but works against you when the design is part of the brand. We make the opposite trade: the design is fixed, the content varies, and the boundary holds across every run.
How it works
A typical setup has three artifacts in play:
- The master template — a Slides/PowerPoint/Word file owned by your design team. It looks like the finished document with placeholder strings standing in for variable content.
- The data source — wherever the variable content lives. See data sources for the full list.
- A field map — created once when you wire a template to a data source. Tells the engine which placeholder corresponds to which data field.
Once those three are in place, generating a new document is a single action. The output is a fresh file in the requested format that matches the master template pixel-for-pixel on the structural elements, with your data injected into the right places.
Designers don’t learn a new tool. They keep working in Slides, PowerPoint, or Word, where they already know the keyboard shortcuts and the limits of the format. The only addition to their workflow is the placeholder convention, which takes about ten minutes to learn.
What’s included
Every workspace gets the full template engine:
- Multiple master templates per project. Build a long-form version and a short-form version of the same report and pick at generation time.
- Conditional sections. Show, hide, or substitute regions based on data values. Useful for “if this account is at risk, show the risk slide; otherwise skip it.”
- Repeating sections. A single master slide produces one slide per item in a list. Useful for speaker bios, account lists, portfolio company updates.
- Brand layers. Override logo, primary color, and font family per project without forking the master template. The same template can serve five clients with five brand identities.
- Image and shape placeholders. Replace any shape in a Slides deck or PowerPoint file with an image fetched from a URL, a data source, or a static asset library.
- Text formatting preservation. Bold, italics, links, and line breaks inside a field are preserved through generation.
Where it fits
A few of the workflows the template engine is built for:
Recurring branded reports across many recipients. A marketing agency producing 50 monthly client reports from the same master template. A customer success team producing 200 QBR decks. A fund producing 25 quarterly LP reports. Same shape, different data, brand survives.
Long-form documents with variable structure. Conference programs where days, stages, and sessions change weekly. Audit reports where the number of findings varies per engagement. Investment memos with conditional sections.
Multi-brand delivery from a single platform. Agencies and consultancies running the same workflow for clients who each need their own brand identity on the output. Brand layer per project; one underlying template.
Honest limits
A few things the template engine doesn’t do:
- Non-supported master formats. Input templates must be authored in Google Slides, PowerPoint, or Word. We don’t accept InDesign, Figma, or Adobe Illustrator masters.
- Animations and transitions are preserved but not generated. If your PowerPoint master has animations, the output keeps them. We don’t add new animations programmatically.
- Very wide tables (more than ~8 columns). Layout in Slides and Word starts to break around that width. The output may render but readability suffers. We recommend a different visualization for data that wide.
- Complex multi-page Word documents with chained text frames, margin notes, or sidebar overlays have rendering edge cases. The 95% case works; the 5% case requires testing per template.
We name these because they matter to people evaluating the platform. The capability boundary is more useful information than a generic “powerful and flexible” claim.
FAQ
Can a non-technical person author the master template?
Yes — the placeholder convention is simple enough that anyone comfortable in Slides or Word can use it. The harder part is usually the field mapping, where someone needs to know which data field corresponds to which placeholder. That step is typically owned by someone with light technical familiarity (an ops manager, not necessarily an engineer).
What happens when the data is longer than the template space allowed?
Text fields auto-wrap. Repeating sections produce as many slides or pages as the data requires. Image placeholders scale to fit their container while preserving aspect ratio. If a designer wants to enforce a maximum (a 3-line bio, a 200-word description), they can — but the default behavior absorbs variable content gracefully.
Can I have multiple variants of the same template?
Yes. You can attach more than one master to a single project. At generation time, the caller specifies which variant to use. This is how teams ship a “full” version and a “summary” version of the same report, or English and a translated variant.
How do template updates flow through?
Update the master file. The next generation uses the updated version. Existing generated outputs aren’t retroactively updated — they’re snapshots of what the template looked like at generation time, which is usually what you want for audit and version tracking.
Does the engine work with my existing brand book?
Whatever brand decisions are baked into the master template — colors, fonts, logo placement, spacing — are preserved on every generation. If your brand book exists as design tokens, the master template is where you encode them once and they’re enforced from then on.