luckpandaTurn your help centre into answers
Your help centre already holds the answers. People still write in, because asking is quicker than searching. Point the chatbot at your articles, documents and FAQs, and it does the searching for them.
No account needed to try it. We read the public pages of your site, build the chatbot in about a minute and delete it after 7 days unless you keep it.
- Up to 50 pages per website source
- PDF, DOCX, TXT, MD and CSV up to 5 MB
- Says when no article covers it
A question your docs answer, and one they do not
Sample conversation on a fictional Wi-Fi router maker's help centre
The second question is a gap in the docs, not a failure. It is now in History, so you know which article to write next.
What it can read, and how much
| Source | Good for | Limits |
|---|---|---|
| Website | A public help centre or docs site | 5, 10, 20 or 50 pages per source, link depth 0 to 3, same site only |
| Documents | Manuals, policy PDFs, exported articles, product lists | PDF, DOCX, TXT, MD or CSV, up to 5 MB each; CSV up to 5,000 rows |
| FAQ | A common question, answered in your exact words | Question up to 120 characters; edit any time |
| Text | Notes that live nowhere else, such as a known issue this week | Up to 50,000 characters per entry |
A bot holds up to 60 sources. For each question it searches all of them and uses the best three passages.
When the wording matters, add it as an FAQ
A help article is written to be read from the top. A customer asks one question. When a rule has to be stated exactly, such as a warranty term or a cancellation window, add it as an FAQ entry or a text answer as well as keeping the article.
The FAQ question works as the entry's title, and words in a title count for more in the search. Phrase the question the way customers ask it, and that entry is likely to come up first.
- FAQ and text entries are ready as soon as you save them; there is no crawl to wait for
- You edit them in place, while website pages are re-crawled and documents are replaced
- Open any source to see exactly what the bot learned from it
More than 50 articles
One website source reads up to 50 pages, starting from the address you give and following links on the same site. A bigger help centre needs more than one source.
- Add another website source that starts at a different section, such as your billing articles. It follows links from there, so some pages may be read by both sources.
- A help centre on its own subdomain, such as help.example.com, is a separate site. Add it as its own source; a crawl of the main site will not follow links into it.
- If your help tool can export articles, upload them as Markdown or PDF files instead.
- Pages behind a login, and pages kept out by robots.txt or noindex, are not read. Each skipped page is listed with the reason.
When an article changes
- Edit the articleUpdate it in your help tool as you normally would.
- Press Re-crawlThe bot re-reads the site. Old pages keep answering until the new crawl finishes, and pages that no longer exist are removed at the end.
- Replace changed filesRemove the old document and upload the new one. Edit FAQ and text entries directly.
- Ask againPut the question to the test chat and check the answer uses the new wording.
Re-crawling is manual today. Scheduled re-crawls are not built yet, so make Re-crawl part of publishing an article.
Files and pages it skips
- Scanned PDFs, which are pictures of text. There is no OCR yet, so the upload fails with a reason. Upload a text-based PDF or paste the text instead.
- Password-protected PDFs, old .doc files, spreadsheets and presentations. Save them as DOCX, CSV or a plain PDF first.
- Screenshots and videos inside articles. Only the text on the page is read, so write important steps out in words.
- Navigation, headers, footers and sidebars. These are stripped so the bot learns the article, not the menu.
Questions
Can it read a help centre hosted by a help-desk tool?
If the help centre is public, yes: it reads it like any other website. There is no direct connection to help-desk tools, and articles that need a login cannot be read.
How often does it re-read my site?
Only when you press Re-crawl. Scheduled re-crawls are not built yet. While a re-crawl runs, the old pages keep answering.
What happens to a scanned PDF?
The upload fails with a reason saying no readable text was found. There is no OCR yet. Upload a text-based PDF, or paste the text as a text entry.
Will it answer questions the docs do not cover?
It is told not to. When the search finds nothing, it says it could not find the answer and offers to pass the chat to a person. It can still be wrong, so test it with real questions before you go live.
Can I see what it learned from a page?
Yes. Open any source to see each page or part it learned, with its text. If a page is missing, the list of skipped pages gives the reason.
Do the limits depend on the plan?
No. Every bot can read 50 pages per website source, hold 60 sources and take documents up to 5 MB. You pay ₹249 a month only to put it on your website.
Related pages
- Customer supportAnswer the questions your site already answers, and hand the rest to a person.
- Multilingual supportYour pages are in English, but your visitors write in Hindi, Hinglish or Spanish. Reply to each in their own language.
- SaaSTrial users ask what your docs already say. Answer from the docs, and send buyers to sales.
- AgenciesQualify enquiries on your own site, and run chatbots for client websites from one account.
Build it with your AI agent
Give an agent your site and it can create the chatbot, read your pages and test it through LuckPanda's API and MCP. It prepares the work; going live stays with you.
Use LuckPanda's MCP server (https://luckpanda.app/mcp) to build a website chatbot for me. Ask for my website address, call chatbot_create_from_website with it, wait until the pages are learned (chatbot_source_list), then run a few test questions with chatbot_test_message and show me the answers. Do not put it live: I will do that myself.
Instructions for AI agents
Needs a LuckPanda agent key (Settings › Developers) with the chatbot.manage grant. MCP endpoint: https://luckpanda.app/mcp (Streamable HTTP, bearer key).
chatbot_create_from_website{ url }: creates a draft chatbot named after the site and starts reading it.chatbot_source_list: pages learned, failed or skipped, with a reason for each.chatbot_test_message: asks the bot a question as the owner. Test chats never email anyone.- Going live (
chatbot_updatewith status live) is a person's action; an agent gets APPROVAL_REQUIRED.
See it on your own website
Paste your address and watch the chatbot learn your pages. No account and no card.