OnBrand by SlideSpeak
Design context for AI agents
Start new thread

OnBrand by SlideSpeak - Design context for AI agents

by•
OnBrand gives AI agents design context through MCP. Add your brand guidelines, logos, colors, icons, imagery and slide rules, so tools like Claude, Codex and ChatGPT can create slides, infographics and design assets that actually match your brand. No more generic AI output, random layouts or off-brand visuals. OnBrand can generate your entire brand guidelines as a design.md from your existing website, existing branding guidelines or even PowerPoint templates.

Add a comment

Replies

Best
Maker
📌

Hello everyone! 👋

Today, we are launching OnBrand, an MCP that infuses any AI agent with your brand guidelines, so you can focus on creating instead of worrying about the agent going rogue.

OnBrand is a new product from the team behind SlideSpeak, a leader in AI presentation generation. At SlideSpeak, we have encountered virtually every use case companies face when integrating AI with their brand. More often than not, AI makes tiny mistakes that, when compounded, end up costing a ton of money and time.

OnBrand solves this by continuously steering the agent to comply with your brand guidelines.

You can try it out today, completely for FREE, at . The only thing you need is a SlideSpeak account. We are also open source:

Let us know your thoughts!

Hey  congrats on the launch! Genuine question, ow are you handling user questions right now, and has that been a pain yet or not really? Just trying to learn what's real this early. No worries if you're heads-down.

I just tried it and I have to say this worked perfectly !

Pairing it with my Claude Code subscription gave me surprisingly good results !

Maker

 Thank you!

Amazing! Much needed

Maker

 Thanks a ton! Feel free to let us know your thoughts if you give it a try. We are actively improving it right now :)

Such a cool idea. Congrats on the launch! 🚀

Maker

 Thank you!

Clean framing, especially building it as MCP context instead of yet another wrapper. Where it gets interesting for me is the 'continuously steering' part, a static in the prompt tends to drift over longer runs. Are you re-injecting the guidelines per step, or is there a validation pass after generation that catches the off-brand bits?

Maker

 The latter, the MCP exposes validation steps to the model!

This is a massive pain point. Ensuring sales and marketing teams actually stick to brand guidelines instead of going rogue with outdated assets is a constant battle. I am curious about the ingestion flow—how does the AI handle conflicting legacy assets? If a user uploads an old deck with a retired logo, does the system flag it automatically? Very good positioning for B2B teams, congrats on the launch!

Keeping slide decks on-brand at scale is genuinely painful for growing teams — every new hire rebrands the template their own way. SlideSpeak solving this with AI auto-correction is smart. Do you support custom brand kits with logo + color palette rules? Congrats on the launch!