See below the answers to the questions that were asked during the Aplysia 3 webinar on the 12th of June 2025:
Are documents only in English? If the doc is uploaded in English, would the chatbot be able to reply accurately if the question is asked in a different language?
Yes, documents are currently only supported in English. However, the chatbot will respond in the language set in your chatbot configuration — and it also dynamically adapts to the guest’s language during the conversation. So even if your knowledge documents are written in English, the chatbot will still respond naturally in other languages.
On our platform, we see the Aplysia 3 option disabled? Does this have an additional cost?
Aplysia 3 will be available to all Pro, Premium, and Enterprise clients at no additional cost.
Do we upload different documents for each property, or one generic one with custom responses?
You can do both. Shared information across all properties should go into a Company Knowledge Document, so you don’t need to maintain duplicate content. Property-specific information should be added to each property's individual knowledge document.
If documents have conflicting info over time, will the bot detect or warn us?
No — Aplysia 3 doesn’t automatically detect or flag contradictory information. It's important to ensure consistency across documents yourself. A few tips:
Organize documents with clear sections and headings.
Use the search function before adding new content to avoid duplication.
That said, we’re working on features to help manage growing content more efficiently — stay tuned!
When will Aplysia 3 be available?
It’s already available via a waiting list: https://www.hijiffy.com/aplysia-3-upgrade!
Can we upload our guest directory or travel guide in PDF?
Aplysia 3 only supports the text-based knowledge document in the console. PDF content must be manually copied and converted into the document.
Can the AI read new info directly from our website? Or do we still update the document manually?
Aplysia 3 does not automatically read or pull content from websites. All updates must be made manually in the knowledge document within the console.
If I have both pre-stay and in-stay chatbots for one property, do they read from the same document?
Yes. All Aplysia 3 chatbots linked to a property use the same property-specific knowledge document. You only need to maintain one document per property.
Will Aplysia detect duplicated information in the document and warn us?
No. Aplysia 3 does not currently detect or flag duplicate content. However, we know this is a growing need — and we may have updates on this soon.
For large hotel groups where the document model doesn’t scale — are there alternative solutions?
Currently, each property has its own knowledge document, and there's one shared company-level document for policies and information that apply across all properties. We recognize the need for scalable solutions and are actively exploring improvements.
About the email channel — how does it work for answers?
At the moment, email automation is not supported.
Can you give the AI chatbot a tone of voice (e.g., formal but funky)?
Not officially. Aplysia 3 doesn’t allow tone configuration settings. However, the way you write your document content can influence the chatbot’s tone — so feel free to write in the style you’d like it to mirror.
I can’t find where to configure topics.
Please follow this step-by-step guide: https://app.arcade.software/share/w3JGShpET69IRdjwBzXF
Please tell us about the new model and channels.
Aplysia 3 is a new chatbot type created by our AI experts, designed to revolutionize user and guest interactions by leveraging advanced Retrieval-Augmented Generation (RAG) technology. This technology retrieves relevant information from structured property documents and then generates accurate and context-aware responses using Generative AI. It’s available across all our automated channels..
If the AI gives a wrong answer, can we see where in the document it got it from?
Not precisely. You can see whether the response was generated from a topic or a document, but not the exact section or line. If you believe the answer contradicts the document content, please report it — our support team may be able to trace it via logs.
Can you summarize the top 3 advantages of Aplysia 3 over the old chatbot? I'm not sure what benefits this will bring.
Improved understanding of guest questions: Aplysia 3 can handle complex and property-specific terminology that Aplysia 2 couldn’t. It understands the actual names of your restaurant, spa, or events — and responds accordingly.
Content flexibility: You’re no longer restricted to FAQ-style content or topics. You can write freely, and the chatbot will still interpret and use your content accurately.
Simpler content management: You don’t need to follow rigid templates. Just write in your own words, and the chatbot will understand it — saving time when maintaining your content.
Does the new dashboard still count chats with no user interaction, affecting Automation Rate and CSAT?
No! Based on your feedback, the new reporting only considers chats where the user actually interacted — giving you a more accurate view of automation and satisfaction metrics.
Can we edit the knowledge document that you generate in the migration?
Yes — absolutely! In fact, we encourage you to review and improve it. Think of it like this: you're writing directly for your guest, and the clearer your content, the better the chatbot will perform.
Here are a few tips to guide your edits:
- Use simple, guest-friendly language: Short sentences work best. Stick to English when writing — the chatbot will handle translations during conversations.
- Organize your content well: Break topics into logical sections and sub-sections. For example, under Check-in, you could include:
2. Early Check-in Policy
3. Online Check-in Options
- Be transparent: If a commonly expected service (like room service or a swimming pool) isn’t available, say so directly. It helps the chatbot manage guest expectations and avoid misunderstandings.
- Use formatting to help, not clutter: Avoid long paragraphs. Use headings (H1, H2, H3), bullet points, and tables — but only when they make the content easier to read. Images won’t be processed, so everything should be in plain text.
- Keep it up to date: Aim to review and update your documents regularly — ideally once a month or after any policy, event, or seasonal change.
- Remember: the more structured and accurate your content is, the better the chatbot’s responses will be.
Will the fallbacks still work?
Yes — fallback behavior remains unchanged. If the chatbot doesn’t find a suitable answer and escalation is possible, it will suggest handing it off to a human agent. All other existing features continue to work — including webchat campaigns, WhatsApp campaigns, flows, and more. Just keep in mind the button behavior caveat: if a pre-existing button was linked to an FAQ, Aplysia 3 will now generate the reply based on the button title, using the document.
What happens to existing campaigns?
Your current campaigns will continue to work. However, button behavior changes slightly with Aplysia 3:
- When a guest clicks a button, the chatbot will generate a reply based on the button title, using information from the knowledge document.
- If you want full control over the chatbot's response to a button, you need to create a custom topic answer and link the button to that topic.
Can you explain again how the answer is generated — from documents and from custom responses?
When a guest sends a message, the Aplysia 3 chatbot uses a multi-step decision-making process to determine the best response. This process combines topic logic with document retrieval to ensure flexibility, accuracy, and control.
Step 1: Topic Matching
The system first checks whether the message matches any trained topic.
If a matching Topic is found and is set to Replace AI-generated Answer, the chatbot immediately replies using the fully configured content (text, images, and/or buttons). The document is not used.
If a matching Topic is found and is set to Support AI-generated Answer, the chatbot continues to Step 2 to retrieve text from the document. It then combines this with the buttons or images configured in the Topic Response.
If a matching Topic exists but has no configured Topic Response, it is ignored and the system moves to Step 2.
If no matching Topic is found, the system proceeds directly to Step 2.
Step 2: Document Retrieval
If no suitable topic was found or if the topic response is in enrichment mode, the chatbot searches the applicable Property Knowledge Document and the Company Knowledge Document for relevant information.
The search uses Retrieval-Augmented Generation (RAG) to locate and compile a response from the live documents.
Only documents marked as Generating live answers are used.
Step 3: Fallback Behaviour
If no matching topic and no relevant document content are found, the chatbot follows fallback procedures:
This may include transferring the conversation to a human agent or displaying a generic fallback message, depending on the configuration.
This three-layer approach ensures:
High accuracy in handling property-specific or branded topics
Consistency and up-to-date responses through document integration
Control and customisability via topic-based overrides
Keep in mind this is a simplified overview of the chatbot's decision-making. In parallel with the core response flow, the system also:
Attempts to trigger any configured conversation flows (e.g., Book a Room).
Suggests relevant topics when appropriate, based on guest intent (Proximity Feature)
Evaluates when to transfer the conversation to a human agent for further support.