Import from Website

Last updated: 6 October 2026

This feature lets you pull your business information from your website (or from a piece of text you already have) instead of writing it from scratch. The system brings everything it finds as a draft; nothing is shown to guests until you approve it. Next to every draft you also see the sentence from your site where the information appeared; information that is not clearly written on the site is never suggested.

What is on the screen#

You reach the page with the Import from website button at the top of the Knowledge Base page (Info Center → Knowledge Base → Import from website). To go back, there is a ← Back to Knowledge Base link at the very top.

From top to bottom:

  1. Header row: Knowledge Base on the left, and on the right the usage badge {count} / {limit} imports this month (1 on Basic, 10 on Pro). The badge turns red when you reach the limit. Below it, the Import from Website heading and a description.
  2. Error box (if any, red): starts with Could not complete:.
  3. Website form: the Your website address box, the This website belongs to me / my business. checkbox, the Start Scan button and a short explanation beside it.
  4. Paste Your Information section: a large text box and an Extract Information button.
  5. Latest import card: the status of the most recent run (progress, error or summary) and warnings.
  6. Business info suggestions (if any): drafts for fixed fields.
  7. Knowledge base drafts (if any): drafts for free-form information entries.
  8. Information not found on your site (if any): the list of fields that could not be found on the site at all.
  9. Scanned pages ({n}) (if any): a collapsible list.

Step by step: Scanning your website#

Example scenario: the website of "Deniz Park Hotel", a 30-room hotel, is www.denizparkotel.com. Instead of typing check-in time, breakfast hours, phone and e-mail one by one, you will pull them from the site.

  1. On the Info Center → Knowledge Base page, click Import from website.
  2. In the Your website address box, type the address. Example: https://www.denizparkotel.com. Required. You may leave out https://; the system adds it. The box may come pre-filled with the address you scanned before or the website address in Business Info.
  3. Tick This website belongs to me / my business. Required. If you do not, you get the error "Sitenin size ait olduğunu onaylayın." (confirm that the site belongs to you).
  4. Press Start Scan.
  5. A message ✓ Scan started. Results will appear on this page. appears at the top. In the Latest import card you see:
    • Scanning site… {n} pages read, then Extracting information… {n}/{total} pages and a progress bar.
    • This can take a few minutes. You can leave this page; the import continues in the background.
  6. The page refreshes itself every few seconds. While a run is in progress, the address box, checkbox and buttons are disabled. When it finishes, the drafts appear.
  7. Review the drafts as described below and approve or reject them.

A scan usually takes 1-5 minutes, depending on the size of the site.

When a website scan is started and the website field in Business Info is empty, it is filled in automatically with the address you entered.

How the scan works#

  • At most 25 pages of your site are scanned.
  • Only pages on the same domain are visited (www and non-www count as the same).
  • The site's robots.txt rules are followed; the sitemap is checked first, then links inside pages.
  • Useful pages are prioritised (FAQ, about, rules, services/facilities, rooms, contact, restaurant, transfer). Pages like privacy, cookies, blog, news, gallery and careers are pushed to the end or skipped; PDFs, images and similar files are not read.
  • If the site has several languages, the version in your business's content language is preferred. If no such version is found or it cannot be opened, you are told in the Warnings box (for example "Sitenin Türkçe sürümü bulunamadı …", the Turkish version was not found); in that case information is translated from the site's own language into drafts.
  • Draft titles and contents are written in your business's content language.
  • Local network and internal addresses (e.g. localhost) are not accessed.

Step by step: Importing by pasting text#

If you do not have a website, have the text of an old site, or your site cannot be scanned (some big hotel chains block automatic scanning), you can paste the text directly: an old website's text, a Word document, a brochure or an e-mail.

  1. Paste the text into the box in the Paste Your Information section. The text must be at least 80 characters. More detail gives better results; there is no upper limit.
  2. Press Extract Information.
  3. The same Latest import card shows progress. Instead of a link, the source reads Manually pasted text.
  4. The same engine, the same evidence check and the same review screen are used. Each suggestion is shown with the matching sentence from your pasted text.

If the text is shorter than 80 characters you get "The pasted text is too short. Please enter at least a few sentences." Pasting shares the same monthly quota as scanning a website.

Field by field#

Start forms#

Field What it does Example / Note
Your website address The home page of the site to scan. Required. https://www.denizparkotel.com
This website belongs to me / my business. Confirms the site is yours. Required. Do not scan someone else's site
Start Scan Starts the scan in the background. Only one run can be active at a time
Paste box The free text to analyse. Required, at least 80 characters. Old site text, brochure, e-mail
Extract Information Starts extracting information from the text. Uses the same quota as a website scan

Which Business Info fields can be suggested#

The system can suggest the fixed fields below; when you approve one, the value is written directly to the Business Info page:

Label Note
Check-in time, Check-out time HH:MM format
24-hour reception, Reception hours If 24-hour reception is approved, the hours range is cleared
Breakfast included in price, Breakfast type, Breakfast hours A single time is written like "from 05:00"; a range like 07:30-10:30
Pets allowed, Pet policy
Free cancellation (hours before arrival), Cancellation policy note Days are converted to hours (1 day = 24 hours)
Airport transfer available, Transfer price note
Total rooms / units
Address, Phone, WhatsApp, E-mail For WhatsApp you always get the warning "Numaranın gerçekten WhatsApp hattı olduğunu doğrulayın." (verify the number is really a WhatsApp line)

Other free-form information (restaurant hours, pool, beach, parking, Wi-Fi, room features, house rules, location, caravan/camping details) comes as Knowledge Base drafts. Room prices are deliberately not suggested.

Reviewing and approving drafts#

Business info suggestions#

Each suggestion is in its own card:

  • At the top the field name (e.g. Check-in time) and a badge: New (the field is currently empty) or Differs from current value (the field already has a value).
  • The Suggested box: you can edit the value here. Yes/no fields show a Yes / No selector.
  • The Current box: the present value in your business info (read-only). An empty yes/no field is not treated as a real value, so it shows —.
  • A warning starting with ⚠, if any.
  • The Text on your site box: the verbatim quote from the site and a Source page link.
  • Approve (green) and Reject buttons.

When you press Approve:

  1. The value is validated. If it is wrong, a red error box appears: "Saat SS:DD biçiminde olmalı (ör. 14:00)." (time must be HH:MM), "Sayı girin." (enter a number), "Geçersiz e-posta." (invalid e-mail), "Geçersiz telefon." (invalid phone), "Metin çok uzun." (text too long; over 500 characters), "Değer boş olamaz." (value cannot be empty).
  2. The value is written to Business Info; the message ✓ Approved and saved. appears.
  3. Other suggestions for the same field are considered void and are rejected automatically.
  4. For note fields such as breakfast type, translations in the other languages are refreshed. A translation you corrected by hand is kept but marked "⚠ Source changed"; in that case you see: "✓ Approved. Your manually written translation of this field was kept, but since the original text changed it is marked "⚠ Source changed" on the Business Info page; please review it."

Reject closes the draft: ✓ Draft rejected.

Business info suggestions cannot be bulk-approved, because they are given to guests as fixed, guaranteed answers; you approve them one by one.

Knowledge base drafts#

Each card has:

  • A badge: New or Updates existing record (you already have an entry with the same title but different content).
  • Title and Content boxes: you can edit them before approving.
  • A Current record content box (for an update): shows the old content.
  • A warning if any, the Text on your site box and the Source page.
  • Approve (for an update Approve and update existing record) and Reject.

When you press Approve, a new entry is added to the Knowledge Base with the Website type and active; for Updates existing record, the existing entry's title, content and source address are replaced. If the title or content is empty you get the error "Başlık ve içerik boş olamaz." (title and content cannot be empty).

Warning examples:

  • Alıntıda geçmeyen ifade: "ücretsiz" — sitenizde doğru olduğunu kontrol edin. (Phrase not in the quote: "free" — check that it is true on your site.) The content makes a claim such as "free", "paid", "included", "all rooms" or "24 hours", but the shown quote does not contain it. Check that your site really says so, and fix the content if needed. (This warning text is shown in Turkish.)
  • Bu konu İşletme Bilgileri'nde ayrı bir alan olarak tutuluyor; misafir cevaplarında o alan esas alınır. (This topic is kept as a separate field in Business Info; that field takes priority in guest answers.) The entry's topic (transfer, pets, cancellation and so on) has its own field in Business Info. You may approve if you want to add detail, but also write the basic information in the matching field.

For topics already covered by fields such as breakfast, check-in/out and room count, no separate Knowledge Base draft is produced. If an entry with the same title and the same content already exists, no draft is produced either.

Bulk approval#

Above the Knowledge Base drafts section, an ✓ Approve {n} drafts without warnings button appears. It adds only new drafts without warnings. Drafts with a warning and drafts that update an existing entry must be approved one by one; the note beside the button says how many are left out. Edits you make in the cards are not included in this button: if you edited a draft, save it first with its own Approve button.

Bulk approval respects your plan limit: it adds as many as the quota allows and the rest stay pending (they are not deleted). In that case you see: "✓ {n} knowledge drafts added. {m} drafts are still pending because your plan limit is reached — bulk-approve them again after upgrading to Pro." If no quota is left at all, you get a red error.

Rejecting the remaining drafts#

The small Reject all remaining drafts link under the Latest import card closes all drafts that are waiting for approval.

Information not found on your site#

When the run finishes, the fields that are empty in Business Info and could not be extracted from the site are listed as labels (Information not found on your site). That means the information is not clearly written on your site; fill it in by hand through the Go to Business Info → link. The list shows only the basic fields (check-in/out, breakfast hours/type, cancellation hours, room count, address, phone, e-mail and similar).

Scanned pages#

Open Scanned pages ({n}) to see the list of pages that were read (title, address and character count).

What happens next#

  • The Business Info values you approved are used immediately in Mira's fixed answers.
  • The Knowledge Base entries you approved become active at once and show up in the Knowledge Base list.
  • Drafts you reject add nothing to the system; to get that information later you would have to run another import.
  • If no drafts exist when the run finishes, you see: No new information could be extracted. Your records may already be up to date, or your site may not contain this kind of information. If all drafts have been handled: No drafts left to review.
  • The summary line reads: {n} pages scanned · {n} drafts awaiting review · {n} approved · {n} rejected.

Monthly quota#

Basic Pro
Imports per month 1 10
  • Website scans and text pastes share the same quota.
  • The quota resets at the start of the calendar month.
  • Every run that starts counts against the quota; a failed run counts too. So be careful when typing the address, especially on Basic.
  • If the address is invalid, you did not tick that the site belongs to you, or a run is already in progress, the run does not start and nothing is deducted.
  • When the quota is full and you press Start Scan / Extract Information, one of these messages appears in the error box: "Basic planda en fazla 1 aylık içe aktarma kullanabilirsiniz. Daha fazlası için Pro'ya geçin." (On Basic you can use at most 1 import per month; upgrade to Pro for more) or on Pro "Pro planda en fazla 10 aylık içe aktarma kullanabilirsiniz." (On Pro you can use at most 10 imports per month).
  • Knowledge Base entry limits are separate (Basic 5, Pro 50); approved new entries count against that limit. Updating an existing entry does not hit the limit.

Errors and fixes#

Errors appear at the top of the page in a red box with the prefix Could not complete:, or in the Latest import card with the prefix Import failed:. The message text itself is shown in Turkish.

Message Meaning Fix
Web sitesi adresi girin. The address box is empty. Enter the address.
Sitenin size ait olduğunu onaylayın. The checkbox is not ticked. Tick the checkbox.
Geçersiz adres. / Yalnızca http ve https adresleri desteklenir. The address is malformed. Write it like https://www.yoursite.com.
Kullanıcı adı/şifre içeren adresler desteklenmez. / Yalnızca standart web portları (80/443) desteklenir. The address contains a username or a special port. Enter the plain home page address.
Bu adrese erişilemez. localhost, a local network or an internal address was entered. Enter the public site address.
Alan adı bulunamadı. The domain does not exist or is misspelled. Check the spelling.
Devam eden bir içe aktarma var. Bitmesini bekleyin. Another run is in progress. Wait for it to finish.
Basic planda en fazla 1 aylık içe aktarma kullanabilirsiniz… The monthly quota is full. Wait for the next month or upgrade to Pro.
Siteye ulaşılamadı: … The site does not open or did not respond. Open the site in a browser to confirm it works, then try again.
Site erişimi reddetti (HTTP 401/403). … The site blocks automatic scanning (as big chain sites do). Use Paste Your Information or enter the data by hand.
Site çok fazla istek aldığını bildirdi (HTTP 429). The site temporarily rate-limited you. Try again in a few minutes.
Site hata döndürdü (HTTP …). The site has an error. Try again once it is fixed.
Adres bir web sayfası döndürmedi. The address is not a web page (a file, etc.). Enter the home page address.
Siteden çok az metin çıkarılabildi. Site içeriğini JavaScript ile yüklüyor olabilir (ör. Wix)… The site loads its text into the page afterwards; such sites are not supported. Copy the text and give it through Paste Your Information.
Siteden okunabilir içerik çıkarılamadı. No meaningful page to read was found. Check the address or paste the text.
İşlem yarıda kaldı (sunucu yeniden başlatılmış olabilir). Lütfen tekrar deneyin. The run made no progress for 10 minutes. Start it again. The new run also counts against the quota.
LLM hız sınırı aşıldı, daha sonra tekrar deneyin. The AI service is busy at that moment. Wait a little and try again.
Taslak bulunamadı ya da zaten işlenmiş. The same draft was already handled in another tab. Refresh the page.
Toplu onaylanabilecek taslak yok. No new draft without warnings is left. Review the rest one by one.

Tips and common mistakes#

  • On Basic you have a single import: first open the site in a browser to confirm it works, and type the address carefully. If the site belongs to a big chain or loads its text with JavaScript, use Paste Your Information from the start.
  • The version in your content language matters. On multilingual sites the language versions may not contain the same information; the system scans the version closest to your content language.
  • Compare every draft with its quote. Reading the quote makes your approval safe.
  • Edit before approving. Fix the draft's title/content, then press Approve; bulk approval does not include edits.
  • When a suggestion conflicts with the current value, look at the Current box. If the current value is a correct value you entered by hand, press Reject.
  • Review Business Info after the scan too. The Information not found on your site list shows what is still missing.
  • Do not leave drafts waiting. When you start a new run, you can no longer see the old drafts on this page.
  • Prices are not imported; enter room prices on the Room Types page.
Still stuck?

Write to us and we will help you set it up.

Or email us directly: support@airoomier.com

Write to us