What is Multi-variant? One Chatbot, Many Locations
Multi-variant (locations, courses, etc) lets one chatbot answer for the right location, course or plan, compare across all, and ask which one.
What is Multi-variant (locations, courses, etc)?
Multi-variant (locations, courses, etc), formerly called Hierarchical RAG, is a Hyperleap add-on for one specific shape of knowledge base: many near-identical sources, such as one page per location, one brochure per course, or one sheet per plan, that differ in only a few details like address, hours, fees or duration. With it, one chatbot picks the right location's (or course's, or plan's) source before answering.
Every Hyperleap chatbot already answers accurately from its knowledge base out of the box. With several similar variants, the hard part isn't finding the answer; it's finding it for the right one.
How It Differs from a Standard Knowledge Base
Standard knowledge base (RAG)
A standard RAG chatbot:
- Takes a user question
- Searches all documents for relevant passages
- Passes the most similar passages to the AI model
- Generates a response from them
This works well for most knowledge bases. Where it gets stuck: when ten branch pages read almost the same, "the most similar passages" can come from the wrong branch, a question across all branches only sees the top few, and a question that doesn't name a branch gets one branch's answer presented as if it applied everywhere.
With the Multi-variant add-on
The chatbot adds a step before answering:
- Looks at the names of your sources and works out which ones the question is about
- Works out whether the question is about one variant, several or all of them, or doesn't say
- Answers from the right source, or from every relevant one for comparison questions
- When the question doesn't say which one and the answer differs between them, asks which one the user means
What It Does
1. Answers from the right variant
"What are the hours at your Plano location?" is answered from the Plano source, not from whichever location page reads most similar.
2. Answers across all variants
"Which of your locations are open on Sunday?" or "Compare the 6-month and 9-month certificate programs" gets context from every relevant source, not just the top two matches.
3. Asks instead of guessing
"What are the fees?" at a school with several programs gets "Which program are you asking about?" instead of one program's fees or a blend of several.
4. Follows the conversation
"…and the Round Rock location?" switches to Round Rock instead of staying on the location discussed earlier.
When You Need It (and When You Don't)
| Your knowledge base | Standard knowledge base | With Multi-variant |
|---|---|---|
| One location, one product | Enough on its own | Not needed |
| Content varies by topic (pricing, policies, FAQs) | Enough on its own | Not needed |
| Many near-identical locations, courses or plans | Can pick the wrong sibling or miss some | Picks the right one, covers all, asks when unclear |
Setting It Up on Hyperleap
The Multi-variant add-on costs $40/mo + 2x credits per request and is available on Pro and Max plans. It is not included in any plan. For businesses with several similar locations, courses or plans:
- Upload one source per variant: a page or document for each location, course or plan
- Name sources clearly: "Plano Location", "Round Rock Location", "9-Month Certificate Program"
- Keep shared information in its own sources: brand policies and general FAQs stay available for every variant
- Turn on the add-on: the chatbot picks the right source for each question, covers all of them for comparison questions, and asks which one when a question doesn't say
Common Questions
Does it require special document formatting?
No. It works from the names of your sources, so clear names ("Austin Office", "San Antonio Office") matter more than formatting.
Does it make answers more accurate in general?
No. Answering from your documents instead of the model's training data is what every Hyperleap chatbot already does. The Multi-variant add-on solves one specific problem: telling near-identical locations, courses or plans apart.
Can it handle multiple products or brands?
Yes. Locations, courses, plans and product lines all work the same way: one source per variant, named clearly.
What about information that applies to every variant?
Keep it in its own sources (for example, "Cancellation Policy"). It's still found for every question, whichever variant the question is about.
Getting Started
- Start a 7-day trial at hyperleap.ai/start
- Upload your documents, one source per location, course or plan
- Configure your AI agent with brand voice
- Deploy across channels (website, WhatsApp, Instagram, Messenger)
- Add the Multi-variant add-on (Pro or Max) when your knowledge base has several similar variants
For a longer walkthrough, read One Chatbot for Many Locations or Courses.
Related Terms
- AI Chatbot for Multi-Location Businesses: The product application — one chatbot answering correctly across every branch
- RAG (Retrieval-Augmented Generation): How every Hyperleap chatbot answers from your documents
- AI Agent: The conversational interface that answers from a RAG knowledge base
- Conversational AI: The broader category of AI-powered conversation systems