Why RG AI Will Make You Bring Your Own Key
Most AI products resell someone else's API at a markup. RG AI takes a different route, and it changes who's actually in control.
Almost every AI chat product works the same way: you pay them a monthly fee, and behind the scenes they call OpenAI, Anthropic, or Google on your behalf, keeping the difference. It's a simple business, but it means your access is only as good as their margin, and your data passes through a server you'll never see.
A different model
RG AI doesn't hold any API keys of its own. When you connect a provider (Claude, ChatGPT, Gemini, Kimi, or DeepSeek), you paste in your own key, and it's stored only in your browser's local storage. Your request passes through a lightweight RG AI route on its way to the provider you chose, but that route doesn't log or store it. Chat also runs free on RG AI's own built-in model any time you'd rather not connect a key at all.
- No subscription markup: you pay the provider's actual rate, not ours.
- No server-side key storage: there's nothing to leak in a breach.
- No vendor lock-in: switch providers any time, per conversation if you want.
The trade-off, honestly
This isn't free of friction. You need to get your own key, and some providers require billing details before they'll issue one, even for their "free" tier. Gemini and DeepSeek are the clean exceptions: both hand out working keys with a genuinely free tier, no card required. That makes one of those two the fastest way to try RG AI on your own key, before deciding whether to connect a paid provider like Claude or ChatGPT for heavier use.
You're not renting access to AI. You already have it. We're building RG AI to give it a better place to live.
Next up
One Workspace, Five Providers: How Model Switching Works