I need to train my bot on about 7000 website pages but it doesnt seem to be able to handle so many site pages. How can I train it on a large website?

I am currently facing problems building my nextJS React application with

@botpress/webchat@3.2.12

The error I am getting while building the app is as follows:

/node_modules/@botpress/webchat/dist/index.js:23718:1
Module not found: Invalid generator object. Asset Modules Plugin has been initialized using a generator object that does not match the API schema.
 - generator has an unknown property 'filename'. These properties are valid:
   object { binary?, dataUrl? }
   -> Generator options for asset/inline modules.
  23716 | }), jY = {
  23717 |   // Optimized OGG file (4.4KB instead of 124KB WAV)
> 23718 |   chatMessageNotification: new URL("data:audio/ogg;base64

There seems to be no documentation on how to build this or what kind of build system is required. I also noticed in the distributed webchat backend that the source code still contains an import to a

style.css

which is not common in a ready to use component package. There is also no documentation anywhere on this. Unfortunately the

@botpress/webchat

also doesn't seem to be open source the repository is pointing to an unreachable location in github. So there is also no easy way to look into this myself. Did anyone else face a similar issue?

Hi, I setup Botpress for a client, they love it, and now they are selling their company and need the email address switched over to a new account. Can you help me switch my accounts current email address to theirs?

Hey there, I'm currently struggeling with adding a FAQ-first feature to my autonomous nodes. The idea is to : - Always search the FAQ knowledge base for things that could slightly affect the response - Always search a second knowledge base that has the "real" data I need these KBs to be seperate, because there might be FAQs like Q:"when doese the skiing resort open?" A:"skiing resort will open on the 04.12.2025" And somebody asks for the lift services. So in the regular RAG it won't show up because it might not match with high enough relevance. Another example could be a QA about a blocked road or something similar. I tried adding 2 search knowledge cards with the different KBs but the LLM never uses both tools, however hard I try with the prompts (also card prompts). Is there a best practice on how to achieve that?

I'm currently working on the bot and I'm having a problem: whenever I test the bot from an emulator it works perfectly, but when I try it from WhatsApp, Telegram, or WebChat it doesn't work.

I have a WhatsApp bot that handles order bookings from stores and restaurants. Therefore, I have an API link containing restaurant information in JSON format. Each object contains some information; for example, a restaurant has things like its name and a link to its menu. The challenge is that the entire restaurant list is contained within a single link, and each restaurant has its own data, such as the menu link. What I want is to create a bot that, if a customer requests a burger from a specific restaurant and specifies their order, will retrieve that information with its details according to my desired format. I've tried many methods to implement this, but I haven't been able to. (Note: I have an API key. I want the bot to retrieve the menu link when the restaurant is called, then retrieve that link to access the menu and answer the customer's queries.) The most difficult aspect I encountered was the input and output commands. I couldn't understand them and tried to have them written by an AI, but sometimes I get results, though the output is often in the form of code like

div a href

. I watched a YouTube video explaining the action, but it wasn't very detailed. i spend more than three days i was working in this. @loud-crowd-60573

I'm trying to use the following code to start a proactive twilio conversation (basically get a conversationId): console.log('Main webhook payload:', event.payload.body) user.customerName = event.payload.body.name user.customerPhone = event.payload.body.phone const result = await actions.twilio.startConversation({ conversation: { userPhone: user.customerPhone, activePhone: env.twilio_number } }) console.log('Generated convId object:', result) event.conversationId = result.conversationId The error received in the logs is: Error while executing integration action: An unexpected error occurred in the integration. Bot owners: Check logs for more informations. What am I doing wrong? If I try to use the Start Proactive Conversation card, I get a slightly different error that says: Error while executing integration action: Could not create conversation: missing channel, channelId or userId. Do I need to create a user using the twilio card? My intent is to start a new conversation with the received context data from the webhook.

???? https://cdn.discordapp.com/attachments/1438436821007597618/1438436821490208768/Screenshot_2025-11-13_at_08.53.53.png?ex=6916e02f&is=69158eaf&hm=15e32057be120069f51cb8a34754cbd6b60d9628f19a9f065407e0cf43e72a43&

[google-ai] Google AI error: {"error":{"code":429,"message":"You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/usage?tab=rate-limit. \n* Quota exceeded for metric: generativelanguage.googleapis.com/generate_requests_per_model_per_day, limit: 0","status":"RESOURCE_EXHAUSTED","details":[{"@type":"type.googleapis.com/google.rpc.Help","links":[{"description":"Learn more about Gemini API quotas","url":"https://ai.google.dev/gemini-api/docs/rate-limits"}]},{"@type":"type.googleapis.com/google.rpc.QuotaFailure","violations":[{"quotaMetric":"generativelanguage.googleapis.com/generate_requests_per_model_per_day","quotaId":"GenerateRequestsPerDayPerProjectPerModel"}]}]}} Is this error message my fault? I only sent a test message with Execute Code. There is no other traffic in the chatbot.

Hey botpress-team, In the Home tab of the studio there is usually a field to edit the general instructions for the chatbot right? I've seen and edited it but it doesn't show up anymore: https://cdn.discordapp.com/attachments/1438207044820795544/1438207045114400999/image.png?ex=69160a30&is=6914b8b0&hm=2d36c735adbc4d9f5e85535213f4f361e01aa07eaf69c461eeed8ea4e6fc4c6e&

Hi Botpress team, I'm trying to build a bot on your platform since a couple weeks and I found a couple issues in the Botpress Studio: Your software is feature-rich and promising but the editing experience is sometimes nerve-wrecking. 1. Backspace deletes entire node without warning It happened to me multiple times that I was working on an autonomous node's instructions. Due to some lag / issue the instructions window lost focus and instead the node gained focus and was immediately deleted because I guess the last input I gave was a backspace. In most cases I was able to recover the node but just now I lost 2 hours of work because the Ctrl+Z wouldn't work anymore. 2. Studio freezes and shows full black screen It happened to me also multiple times already that the studio freezes completely and I see in the Console that message. Only a page refresh solves the issue and some edits were lost. Console: vendor-7bcd6c2.js:50852 Error: Minified React error #300; visit https://reactjs.org/docs/error-decoder.html?invariant=300 for the full message or use the non-minified dev environment for full errors and additional helpful warnings. at Xh$4 (vendor-7bcd6c2.js:64515:13) at dj$1 (vendor-7bcd6c2.js:65218:42) at Wk$5 (vendor-7bcd6c2.js:67316:88) at Vk$1 (vendor-7bcd6c2.js:67018:13) at Uk$5 (vendor-7bcd6c2.js:67009:7) at Jk$1 (vendor-7bcd6c2.js:66996:9) at Ok$4 (vendor-7bcd6c2.js:66759:82) at Fk$4 (vendor-7bcd6c2.js:66815:30) at jg$1 (vendor-7bcd6c2.js:63725:17) at vendor-7bcd6c2.js:66653:24 3. Window "updated" / "refreshes" by itself Once in a while the botpress studio rerenders itself completely and therefore loosing focus of the current editable field. I think it's a react state change issue you are having there. Also these re-renders make the editing experience quite a balancing act. I started to write all prompts and instructions in local files again, because editing live is too dangerous.

Hi Botpress team, quick sanity check on AI Spend. Plan: Pay-as-you-go Workspace: Digital Palestine by SSA Period: last 30 days Usage: 12 sessions, 63 bot msgs, 33 user msgs, 7 users Dashboard: AI Spend shows $6.47 of $25 used (screenshot attached) This feels high for a few light tests. Logs show most turns at ~5k–7k tokens costing ~$0.012 on gpt-4.1, but I also see spikes around $0.12 that look like a higher-tier call or multi-iteration. Questions: 1) Is this spend normal for this volume? 2) How do I hard-lock the bot to a single low-cost model, e.g., 4o-mini or 4.1-mini? 3) How can I disable improvement iterations and any dual-model fallback in Autonomous nodes? 4) Best way to cap cost per turn and avoid long reasoning chains? 5) Any caching or KB lookup settings to reduce repeated tokens? Thanks!

How can i still have a fast replying bot with these many pages. i feel like the bots get super slow after 50+ which is understandable. Maybe its possible to compromise the sites to make them a small description of the sites and upload them as files. Anyone has any experience with this?`let me know

hello, i was trying to import JSON of restored bot state into my bot but i could find only import but that doesnt work. Nothing happens, im not sure how to import this JSON. Any help please ? Thank you very much

I'm hoping this is a softball question for someone. I'm brand new to the platform. I've got the AirTable integration set up and it shows up with available actions for cards. But I'm not able to figure out how to use its records as a knowledge source. Should it show up like Notion does? Or do I have to integrate it with a Botpress table? I'm trying to set this up for a client to be able to administer the knowledge within AirTable, as they are comfortable with that platform. Thanks for any help you can provide!

As the title says... I click the 'EDIT (Search)' function on the left side, below 'Bot Settings'... and NOTHING. Does NOT work. I've restarted my PC, browser, and logged out and back in. SAME. 'ctrl /' does NOT work either. Please advise/fix.

Hi guys, So ive been heavily testing my bot today (webchat), and now it will barely respond, it takes ages and especially triggers. i keep getting this in my logs:

Error sending outgoing message {
  integration: 'webchat',
  messageType: 'text',
  payload: '{"type":"text","text":"Applying settings..."}',
  error: TimeoutError: Sending message took more than 10s
      at Timeout._onTimeout (/var/task/customer_code.js:159:8932)
      at listOnTimeout (node:internal/timers:581:17)
      at process.processTimers (node:internal/timers:519:7)
}

is any one else experiencing this latency in their bots today? https://cdn.discordapp.com/attachments/1437814309903143012/1437814310192807936/Screenshot_2025-11-11_at_14.40.08.png?ex=69149c6c&is=69134aec&hm=1c3646e2fcd976285ab7324c50a2e547b70b6b980533aa59711bce8eab4b821e&

Hey, Botbuilders! I am having trouble copying text into my my clipboard, it ends up copying hidden information (check video)🆙 https://cdn.discordapp.com/attachments/1437804437832728748/1437804438273261598/Recording_2025-11-11_145855.mp4?ex=6914933b&is=691341bb&hm=42e1615229094a7dc7bb7cf4d05fc9a0c6cca4a32e09b750c39506e5ded1ef82&

Hi, fellows! When I try to "Invite member" to my default pay-as-you-go workspace from settings / members, I got a "Unexpected Error Occured" screen just after press the button, witout any request to the server. Then the "back to home" button fails too, I have to reload the page to continue navigation. Try to do many times today with four diferent browsers, and always got the error screen. Any suggestion? Thanks! https://cdn.discordapp.com/attachments/1437783446129868911/1437783446335525026/botpress-invite-member-error.jpg?ex=69147fae&is=69132e2e&hm=7173b9507b872b9ddb98e7bed89fe7c1c405c56a747a3b11649fef94b4f7f30c&

I was informed that the userKey is the CRM key and the field used to search for stored user and conversation variables. Could you please provide an example implementation, some code, or any documentation showing how to use this in a new conversation for a returning user?

We can't use our deployed bot because we are being told that we hit our limit of 5k. This is false because we have only sent 30-40 messages cumulatively over the last 30 days as indicated by Botpress' own dashbaord. Is there any way to resolve this ourselves? We are losing valuable time over this.

I'm facing issue with my Robo Mike website chatbot. Previously, I had created a Robo Mike Instagram chatbot, and it's Autonomous node transition is working. The website chatbot is the replica of the Instagram chatbot, but in this (website chatbot) autonomous node transition is not working. The chatbot responds to the user, but the transition does not occur. Instead, it keeps waiting for further user responses, continuously interacting without moving to the next node. Could you please help me to resolve this issue? Reference: I’ve attached a Loom video showing the issue — https://www.loom.com/share/2245e93343ef4565b26c95affa291df6

I'm looking for the best practice for a specific design pattern in Botpress and am very open to any advice! My Current Architecture: - I have a Main Orchestrator built as an autonomous node. It manages the entire conversation by calling various Expert Sub-workflows (like Expert_Product, Expert_Workflow, etc.) based on its instructions. - The Orchestrator can call multiple Experts for a single complex query. It is responsible for synthesizing all their responses into one smooth, cohesive answer. Goal & Challenge: - I want to conditionally send an image for specific inquiry (e.g., product accuracy), but I'm facing several challenges with this architecture. - The Expert Sub-workflow can't send an image directly with image card because the final synthesized text might be a blend of multiple topics. - The image must be sent after the main synthesized text response because my Main Orchestrator does the synthesis at the very end. The Decision Logic: The decision of which image to provide (if any) is that it needs to be based on both the initial user input query and the content of the final synthesized answer. I'm stuck on how to build this logi and need some advice how to implement such workflow.

Since upgrading from Webchat v1, my bot's responses do not appear in the chat widget. Looking at the logs, I can see the bot is receiving messages, processing them correctly, and sending a response. But the response doesn't appear on the client. If I completely refresh the page, the responses DO then appear, so it seems to be a problem with the way the page receives new messages from the server. This happens on my website, but also with the bot's shareable link, so I don't think it's a problem with the way I'm referencing and initializing Webchat. I've also replicated this issue on a newly created bot that simply responds with the same message when it receives anything. Does anyone know what might cause this?

Hello! I have an upcoming event this week where I need to deploy a KB based chatbot to answer questions from it. I have no experience on what this requires resource wise, do I need to buy ai tokens, upgrade the number of messages per month? (I am still in basic tier) Please help!! Also if you have code to limit the questions to 3-4 per email entry please let me know 😄

Hi there! I would like to change the ownership of our botpress workspace to another account, that is already a member with the role "administrator". At the moment i got the ownership. In some other issues I've seen that this is only possible with the help of community support. Thanks!

Hi, is there a way to export all conversations or, even better, to retrieve them using an API and always get the latest ones to process? I know there is the chat API, but that only helps me directly in the node, and only if I store user variables and not afterwards.

Hi everyone, is it possible to use Zoho SalesIQ as HITL without the Plus Plan ?

I noticed in my workspace that the variables often appear as a normal text instead of a "green bolded text". Although I have already enabled them at the bottom for read and write but in the instruction column, it appears as a normal text. It does not happen all the time but it will happen. And I'm worried that it is not reading the variables https://cdn.discordapp.com/attachments/1437370629958664262/1437370630314917979/Screenshot_2025-11-10_at_17.07.32.png?ex=6912ff37&is=6911adb7&hm=b168a4a67ab2b4b33dc490a7eb31d0e5a1d9d563e61b3c86b0dd8dc1996e120d& https://cdn.discordapp.com/attachments/1437370629958664262/1437370630818369639/Screenshot_2025-11-10_at_17.08.36.png?ex=6912ff37&is=6911adb7&hm=a8db682febe596f6d595e8a18f2d3ddcbc407daf9945f5ddee882f4f0749bdb0&

hi all. This is Danie from a vet clinic. I have almost completed a chatbot that can answer most daily inquiries. But when we want to move to the next steps, which is trying to connect to the ezyvet online booking system, I have no clue about that. Have you had a similar experience?