Search through the documentation for the Storefront Reference Architecture (SFRA) using AI.
Setting the 'subject' is very helpful when you ask about something that has a different answer in SFRA vs OCAPI
For example, how do you create a new basket using code? If you want the right answer to that question you'll have to set a subject first.
Your browser url will update when you set a subject, saving your decision for the future.
The 'model' refers to the specific AI we use to power your search.
Your browser url will update when you set a model, saving your decision for the future.
Search Only
Use this when you want to search relevant SFCC docs and you don't care about an answer from the AI.
This setting is useful for SFCC experts who need to quickly find relevant docs but don't need concepts explained to them.
GPT-4 Turbo
The newest AI model from Chat-GPT is both fast and intelligent. We use it by default for search queries
GPT 3.5
A wonderful workhorse model. It's very fast, very cheap, pretty smart, and great for simple questions.
Claude Opus
Considered the best AI model by some practitioners. However, it's more expensive than other options.
Claude Haiku
The Anthropic equivalent of GPT 3.5. Excellent at data analysis and following instructions, but worse than GPT-3.5 at multi-turn conversations.
Gemini 1.5 Pro
The best Google AI. Probably better than GPT-4, and has the largest context window out of all available models, making it able to remember large amounts of information.
Gemini 1.0 Pro
The Gemini equivalent of GPT-3.5, a cheap workhorse model.
Use this form to suggest an improvement like a new feature, a change to an existing feature, or a new source of docs.
Your suggestion has been successfully submitted. Thank you!
You can use premium AI models for free through BridgeGPT within limits.
The limits reset each day.
This is how much each AI model costs:
At the moment there is only a free version of BridgeGPT. However, if you want to request a paid version that has no usage caps you can reach out to: the Creator.
This represents the likelihood that your answer is accurate and free of hallucinations.
We always recommend double-checking the provided sources, even if you receive a high confidence answer.
If you receive a low confidence score, you might want to rephrase your question and try again.