# How to brand documents with a logo watermark using AI Add a logo or stamp as an image watermark with an AI agent over MCP, with opacity and rotation control, across documents and images. A logo watermark says where a document came from without anyone having to read a footer. [`add_image_watermark`]({{< ref "watermark/mcp/tools-reference/add-image-watermark.md" >}}) places it, and `opacity` decides whether the result is branding or a mess. {{< alert style="info" >}} The commands and config snippets on this page are for the **.NET** build of the server — the only platform available today. Installation and client setup: [MCP server for .NET]({{< ref "watermark/net/mcp/_index.md" >}}). Other platforms will expose the same tools with their own launch command; everything else on this page applies unchanged. {{< /alert >}} ## The prompt > Put logo.png on every page of report.pdf at 30% opacity. Both the document and the logo are resolved by name from your storage folder, so keep the logo next to the documents — or in the folder you configured — and refer to it by file name. ## Getting it to look right * **Opacity 0.2-0.4** — visible, still readable underneath. This is the range most people want. * **Opacity above 0.6** — the logo dominates; useful for a specimen or a sample marking, distracting for a document people must read. * **Rotation** — 0 for a clean centred mark, `-45` for the diagonal look that matches text watermarks. Iterate in the conversation rather than guessing: > That is too strong — redo it at 0.2. Each attempt writes a new file, so the original is never at risk; delete the rejected copies afterwards. ## Logo and text together Brand *and* state the status: > Add the logo at 25% opacity, then add a CONFIDENTIAL text watermark to that result. Two calls, chained on the produced file — the "to that result" is what keeps both marks. ## Practical notes * **Use a transparent PNG** where possible: a logo on a white rectangle will show that rectangle over the page. * **Opacity and rotation only.** [`add_image_watermark`]({{< ref "watermark/mcp/tools-reference/add-image-watermark.md" >}}) has no position, size, or margin parameter. * **Images are targets too.** Branding a set of product photos works exactly like branding a PDF. * **Marks can be removed later.** Because the logo is applied as a watermark object, [`remove_watermarks`]({{< ref "watermark/mcp/tools-reference/remove-watermarks.md" >}}) can take it back off when the document is rebranded — one of the practical advantages over pasting a logo into the source file. Its `textFilter` matches watermark text only, so removing an image logo removes every watermark on the document, text marks included — search first. * **Check the licence before a brand rollout.** Evaluation-mode output is not distributable: [`get_license_status`]({{< ref "watermark/mcp/tools-reference/get-license-status.md" >}}). ## Read more * [Enforce brand guidelines on outgoing documents with automated AI workflows using MCP](https://blog.groupdocs.com/watermark/enforce-brand-guidelines-logo-watermark-mcp/) — the same scenario on the GroupDocs blog, with worked examples and the trade-offs.