ShareYourAI Help Center

Who Is It For? How Does It Work?

Build your own AI. Share it with anyone via a single link.

ShareYourAI turns your documents (product manuals, FAQs, training materials, proposals, etc.) into a custom AI assistant. This AI answers questions with your knowledge first, and honestly says so when it can't find an answer. Best of all, just share a link and anyone can chat with your AI, no sign-up required.

Five Use Cases

How Does It Work?

Quick Start: 5 Steps

Start from scratch — log in, create an AI, upload knowledge files, start chatting, and share with others in just 5 steps.

AI Management

Create, configure, and customize your AI to fit your business needs.

AI Overview Page

Click any AI on the "My AIs" page (or its name in the left sidebar) to open that AI's Overview page. At the top are its avatar, name, status and main button, with a gear for AI settings; below are your share pages, the "What your share page looks like" card, and "Its tools", with Developer tools at the very bottom:

AI Character System

Every AI can have its own character. The system provides 4 preset characters, and you can add your own or let AI generate one automatically.

AI Settings Panel

AI settings live on two pages, both opened from the AI Overview page:

1. Basic Info — Open it with the gear (AI settings) next to the AI's name: name, description, avatar, icon, showcase; you can delete the AI at the bottom
2. Role page — Open it with "Role" under "Its tools": the current role, its settings (four persona groups), switching or adding roles (including Let AI write it), and how it chats with visitors (answer scope, suggested questions)
3. Role page: what it always says, and stickers — Further down the same page: "What it always says" (opening line, when it can't find an answer, when you need to step in, goodbye) and "Stickers"

AI-Generate Character

Describe the character you want in one sentence, and AI will automatically generate a complete character profile (name, personality, speaking style, catchphrases, etc.) based on your knowledge base content. You can fine-tune each field after generation.

How to use: on the Role page, click "New role" under "Switch role" → pick a depth → choose "Let AI write it" and enter a description → preview the result → click "Add role", then "Save changes" at the bottom.

Character Settings

Each role's settings are grouped into four cards under "Its settings" on the Role page — click a card to edit that group:

• Basic Info: Character name, identity, tone style
• Character & Personality: Background story, personality traits, speaking style, catchphrases
• Behavior Limits: Taboo topics, forbidden words, honest boundaries
• Cognitive Layer: Core beliefs, formative stories, internal tensions (upgrades the AI from "sounds like a person" to "thinks like a person")

Cognitive Layer: Make the Character Actually Think

Standard persona fields make the AI sound human, but it still falls back to generic answers when asked for an opinion or stance. The Cognitive Layer is the character's inner mind — fill these three fields and the AI will reason from its own beliefs, memories, and unresolved tensions:

• Core Beliefs: 2-3 things this character believes most deeply. Write views that are exclusive to this role — if you could swap them into a different character, they weren't specific enough. E.g. a barista's "good coffee needs no explanation."
• Formative Stories: 1-2 pivotal moments that shaped the character, with concrete details (time, place, feelings). The character will volunteer these stories when relevant topics come up.
• Internal Tensions: An unresolved contradiction the character carries, phrased as "on one hand… on the other hand…". Real people don't have everything figured out — this tension gives the character depth.

Copy / Paste Character Settings

The "More" menu on the current-role card at the top of the Role page has "Copy settings" and "Paste settings" — move a carefully-tuned role into notes, communities, or another AI:

• Copy Settings: One click copies all current character fields (including the cognitive layer) to the clipboard. The first time you copy, a reminder dialog appears — please verify there are no real names, contact details, or personal info before sharing.
• Paste Settings: Reads from the clipboard and auto-fills the fields. Accepts a complete character (overwrites everything) or just partial fields (only updates what's present, leaves the rest).

Answer Scope

Control whether the AI only discusses topics within its role. This affects how the AI responds to questions outside its knowledge area.

Response Templates

Response templates let the AI automatically use preset reply formats in specific situations, ensuring a consistent user experience:

Built-in skills: contacts, reservations and link cards

On the AI Overview page, click "Skills" under "Its tools" to open the AI Skills page. On the skill wall, "Leave contact info" and "Booking" (tagged "Official") and "Link cards" (tagged "Built-in") are three skills ready to use (the old "AI Actions" page has been merged into here). Turn one on with the switch at the bottom of its tile; when visitors mention a relevant need, the AI attaches the matching card below its reply and visitors finish with one tap — no need to watch over the chat. On Pro and up you can open "Leave contact info" or "Booking" and edit its flow directly (for example, ask a few more questions); if an edit goes wrong, use "Restore official version".

How to Set Up

1. Open the AI Overview page, click "Skills" under "Its tools" to open the AI Skills page, find "Leave contact info", "Booking" or "Link cards" on the skill wall and turn it on with the switch at the bottom
2. To change which details are asked or the booking method: on Pro and up, open "Leave contact info" or "Booking" and use the gear on that step of the flow (on Free, tick the fields right on the "Leave contact info" page); for link cards, open the tile and press "Set up" to go to the Link Cards page.
3. Contact fields and the booking method inside a step's gear take effect when you press "Apply" (they apply to the whole AI); other changes in the skill editor are saved from the save bar at the bottom. Link cards live on their own page, and saving a card takes effect right away; click "AI Skills" in the breadcrumb at the top to return to the skill wall

Export contacts & bookings to a spreadsheet

The "Contacts & bookings" section of "Visitor records" in the left sidebar lists every contact and booking visitors leave. Use "Export" on the right of the section bar (Pro plans and above): "Download CSV" gets the list as it is now (opens straight in Excel); for automatic updates, choose "Sync to a spreadsheet" to create a read-only data link. Paste the link into Excel or Google Sheets and the contact cards and bookings this AI receives (from built-in skills and your own skills alike, except sign-ups, which you download from the "Sign-ups" section of the same page) will sync in automatically — no more manual copying. A JSON link is also provided for your own system or automation tools like Make and Zapier to poll. When viewing "All AIs", Export asks you to pick one AI first (each AI exports separately).

Connecting Google Sheets

1. In "Visitor records" in the left sidebar, open "Contacts & bookings", click "Export" → "Sync to a spreadsheet", click "Create export link" in the window that opens, and copy the complete formula on the "Google Sheets" tab (date/time formatting is built in — nothing to edit)
2. Open a blank sheet tab and paste the formula into cell A1
3. Done — the sheet auto-refreshes about once an hour and fetches fresh data whenever you open it; for every-minute sync, use "Every-minute auto sync" below instead (pick one or the other). After a downgrade to the free plan the link keeps working for 30 more days, then pauses and shows a note in the sheet; once your plan is back, the sheet recovers within about an hour — or click "Regenerate" under "Export › Sync to a spreadsheet" in "Visitor records" and paste the new link if you would rather not wait.

Want every-minute auto sync? (Advanced, one-time setup)

Google caps formula-based refresh at roughly one hour (an official limit that can't be sped up), so every-minute sync needs a small script. It's already prepared with your link baked in, in the "Sync to a spreadsheet" window — five steps, once:

1. In the "Sync to a spreadsheet" window, expand "Want every-minute sync? (advanced)" on the "Google Sheets" tab and copy the script
2. In your sheet, open Extensions → Apps Script, replace everything in the editor with the pasted script, and save
3. In the function dropdown at the top of the editor, pick "install", then click "Run" next to it
4. Authorize once when prompted: choose your Google account, click "Advanced" → "Go to project" when the "not verified by Google" warning appears, then tick the permission checkboxes and press "Continue" (the script only reads your own export link)
5. Done — data syncs into the "預約資料" tab every minute with dates and times shown as-is; once the script is in place you no longer need the A1 formula

Connecting Excel

1. In the "Sync to a spreadsheet" window, switch to the "Excel" tab and copy the URL
2. In Excel, go to Data → From Web, paste the link, and click Load
3. Click any cell in the loaded table, then on the Data ribbon open the dropdown next to "Refresh All" → "Connection Properties", and enable "Refresh every N minutes" (set 1) and "Refresh data when opening the file" — fully automatic from then on If the export gains a new column later and it doesn't show up, Excel has pinned the column count: delete the old query and load it again from Web, or right-click the query and choose Load To a new worksheet.

Want your own columns (like "Deposit paid" or "Assignee")?

Two approaches, depending on whether you already have your own table:

• No table of your own yet: treat the export tab as your table and write your own columns in the blank space to the right of the data. Each sync only updates the system's columns and never touches yours, and the row order is fixed — new bookings only append at the bottom and cancelled ones never disappear (their status just changes) — so every row always stays the same booking and your notes never end up on the wrong line. One more tip: we may add new columns at the right edge of the data in the future, so leave at least one empty column between the data and your own columns (or start further right) so a new column never lands on top of yours.
• Already have your own table: treat the export tab purely as a data source and pull data into your table with formulas — grow the booking-ID column automatically with =ARRAYFORMULA('Data tab'!A2:A) (in Excel 365, =DataTab[ID]), fetch the fields you need with XLOOKUP keyed on that ID, and keep your own columns in your own table with full control over the layout.

AI Skills (things your AI does in order)

The AI Skills page lets your AI go through a skill lineup you set up, item by item, when a certain situation comes up. Write “when to use it”, then drag the skills you need from the skill library on the left into the Skill lineup and put them in order; each item can have one sentence telling the AI what to do. The order is the default — for example, if the visitor mentions their budget right away, the AI won't ask again, and if the visitor changes their mind, the AI goes back. The three built-in skills — Leave contact info, Booking and Link cards — sit on the same wall: you can view them, or make a copy to follow your own flow. Adding your own skills is available on Pro and above.

Three ways to start a new skill

1. From an example: recommend options then take contact details, get the details then pass to a person, introduce then help book, quick quote then leave details, dice scoring, passphrase for a reward, assess and give tips, instant alert for decisions, ask visitors to upload files — the lineup is already set; rename it and adjust the details
2. Start from scratch with AI: describe in one sentence what situation the visitor is in and what the AI should do; the AI uses your knowledge base to name the skill, write when to use it, and pick and order skills from the library, then walks you through a review before you create it; if part of it isn't possible with the skill library yet, the AI tells you and suggests an alternative instead of pretending
3. Start from scratch on your own: fill in the name and when to use it, drag skills in from the library on the left and order them (on a phone, tap “+ Add from skill library” and use the up/down arrows). If you're not sure what to pick or how to order it, tell the “Skill assistant” above the lineup what you want and let it arrange it. Then review and create

The gear on each item in the lineup holds that item's settings (such as which cards to pick from or what else to ask); the defaults work without opening it. A new skill is created when you tap “Create skill”; turn it on and your AI starts using it. “Chat with your AI” on the right lets you test it right away and shows which item the AI is on. The “Skill assistant” above the lineup is a back-and-forth chat: ask it to change something (for example, “After a correct answer, show a discount card, then ask for contact details” or “Add scraping for 500”, then “Cupping too, 600”), or ask about the current settings (for example, “How much is the 60-minute foot massage now?” or “Why can't visitors use this?”). If something is unclear it asks you first. It only changes what you mention; other steps and settings stay as they are, and it can't see or change passcode answers and rewards. Changes are listed one by one first and nothing changes until you click Apply; after applying, the step cards show how many changes were just made. The conversation stays on this page only. Each message counts once and shares its limits with “Write it with AI” (an hourly and a daily limit; the daily one resets the next day).

Notifications

Whether you get an email when visitors pass details or files to you is set under Notifications at the top of the AI Skills page (the same Notifications button appears in "Visitor records" in the left sidebar once you pick an AI): one row for General contact details (left without using a skill) and one row for each skill that collects things, each set to right away (one email each time), daily digest (one email early each morning listing anything not sent yet, up to 50 contacts and 50 submissions per email with the rest in the next one), or don't email (check "Contacts & bookings" and "Inbox" under "Visitor records" in the left sidebar). When one visitor uses several skills at once, you're notified the fastest way among them, in a single email; skills you add later start with the General contact details row. Bookings sent through System email booking with a valid visitor email always email you right away, since you confirm them from a link in that email; bookings without an email follow your list.

Sticker System

ShareYourAI offers an AI sticker response feature that lets your AI automatically send expressive stickers during conversations, adding emotional interaction and reducing perceived wait times.

How Do Stickers Work?

When a visitor sends a message, the AI automatically decides whether to respond with a sticker:

1. Instant Emotion Reading — AI analyzes the visitor's emotion and intent within 0.5 seconds
2. Sticker Selection — Matches the best sticker based on emotion (happy, grateful, encouraging, thinking, sorry, etc.)
3. Full Response — After the sticker, the AI provides its complete answer
4. Detailed Answer — Then the AI provides the full knowledge-based answer

When Are Stickers Sent?

The AI automatically decides when to send stickers based on conversation context. Below are some common examples, but stickers are not limited to these scenarios — each sticker has its own emotion tags and context descriptions that the AI matches flexibly:

These are just common examples. When you upload custom stickers, the system automatically analyzes their emotion and context. You can also manually edit descriptions to define when they're used — for example, upload a "cheer up" sticker with the description "use when the visitor expresses confusion or encounters difficulty, to encourage them", and the AI will send it in matching conversations.

How to Enable Stickers

1. Open the AI Overview page, click "Role" under "Its tools", and scroll to "Stickers" at the bottom of the Role page
2. Click "Manage stickers" and turn on "AI stickers" (takes effect right away)
3. Optionally manage which individual stickers are enabled or disabled

Sticker Sources

Upload Custom Stickers

You can upload your own stickers so the AI uses expressions that match your brand:

1. In "Manage stickers", click "Upload sticker"
2. Select an image (PNG, JPG, GIF, or WebP supported, max 2 MB)
3. The system automatically analyzes the sticker's emotion and suitable contexts
4. You can manually edit the sticker's description and category

Individual Control

• Toggle each sticker (system or custom) on/off individually
• Bulk enable or disable all system stickers / all custom stickers
• Disabled stickers will not appear in AI conversations

Sticker Best Practices

• Shop front desk: Recommended — more human
• Expert Advisor: Selectively enable a few composed stickers to avoid being overly playful
• Role Play: Use custom stickers that match the character's style for deeper immersion
• Branding: Upload brand mascot or company-style stickers to reinforce brand identity

Document Management

Learn how to upload files for your AI to learn from, and how to make AI answers more accurate.

Why Upload Files?

The AI needs to "learn" your data before it can answer questions related to your business. Think of the AI as a new employee — you give them your company documents and manuals, and they use that knowledge to answer customer questions. Without uploaded files, the AI can only answer general questions and cannot provide answers about your products, services, or internal knowledge.

Supported File Formats

Document Processing Status

After uploading, the system processes files automatically. Each file card on the Knowledge page shows its status:

Improving AI Answer Quality

1. Ensure relevant info exists in your knowledge files — The AI can only answer based on what's in your files — if it's not there, the AI can't answer.
2. Organize files in Q&A format — Writing FAQs directly in your files helps the AI find matching answers more accurately.
3. One AI per topic — An AI with a clear focus finds the right information more easily and avoids confusion.
4. Use response templates — Enable the "not found" template so the AI replies with your fixed wording when it can't find an answer, keeping responses consistent.

Q&A Format Example

Here's an example of a well-formatted Q&A document:

Q: What is your return policy?
A: We offer a 30-day unconditional return policy. As long as the item
   is unopened and in good condition, you can request a full refund.
   Please bring your receipt to any of our stores.

Q: How is shipping calculated?
A: Orders over $50 qualify for free shipping. Orders under $50
   have a flat $5.99 shipping fee. International shipping varies.

Q: Do you offer installment payments?
A: Yes. We support 3, 6, and 12-month interest-free installments
   for Visa, MasterCard, and JCB credit cards.

What If AI Answers Are Inaccurate?

If you find the AI's answers aren't accurate enough, try these approaches:

1. Check file content — Make sure your files actually contain the information you expect
2. Improve file format — Use clear headings and paragraph breaks to structure information
3. Add Q&A files — Create dedicated Q&A files for frequently asked questions
4. Use response templates — Make sure the "not found" response template is enabled so no-answer replies stay consistent
5. Re-upload files — If you've updated file content, delete the old file and re-upload so the AI re-learns

Image Content Recognition (Pro and Business plans)

A lot of valuable information isn't in text files — menu photos, posters, product shots, scanned contracts, or screenshots embedded in PDFs and slides. The Pro and Business plans let the AI answer questions based on what's in those images.

What's covered?

Not just standalone image uploads — images embedded in regular PDFs and slides count too:

• Image files (PNG / JPG / WebP)
• Scanned PDFs (paper scans, image-only PDFs)
• Screenshots, product shots, and diagrams inside regular PDFs and PPTX

What is it good for?

• Restaurant menu photos — Upload a menu photo and the AI can answer "What's in the signature set? How much is it?"
• Paper contract scans — Upload the scanned PDF directly and the AI can look up specific clauses
• Product catalogs (mixed text + photos) — Upload a PDF / PPTX with product shots — the AI can answer both specs and appearance
• Tutorial decks & software walkthroughs — Upload a PPTX with screenshots and the AI can explain what each screen does

Important notes

• Monthly quota: 200 pages/images on Pro, 1,000 on Business — exceeding it will block further uploads. Pages count once they're recognized: deleting the file doesn't give them back, and blank pages count as one page each
• Clearer image, larger font, and stronger contrast produce better results
• Printed text works well; messy handwriting may not be recognized accurately
• Pure graphics, logos, and scenery photos with no readable info aren't worth uploading on their own — they just waste quota

How to use it

Exactly the same as regular files — just upload and it's handled automatically. If your plan doesn't include this feature, the upload window says photos and scanned documents need the Pro plan, and images aren't listed among the formats, so you don't upload them by mistake.

Live Cloud Files (Pro / Business)

Regular uploads are a snapshot — once uploaded, the content is frozen, even if the original file changes. Live Cloud Files connect directly to your files in Google Drive: change one word in the cloud document, and the AI answers from the new version immediately — no re-upload needed. Keep in mind that live data is visible to every visitor who asks, so only include content that can be public (remaining slots per time, menus, stock) — never customer names, phone numbers, or booking details. To let the AI answer "which slots are still open", feed it an availability sheet, not your booking list.

What is it good for?

Best for short, frequently-updated content. For large, stable knowledge (product manuals, SOPs, course materials), regular file uploads still work better.

• Use Google Sheet for — inventory, remaining slots per time / scheduling, promo lists, menus (with daily specials), class schedules
• Use Google Doc for — company announcements, latest news, weekly/monthly reports, event agendas

Supported cloud formats

• Google Sheet: Google Spreadsheets
• Google Doc: Google Docs
• Microsoft Excel (.xlsx): .xlsx files stored in Google Drive
• Microsoft Word (.docx): .docx files stored in Google Drive

Authorization & security

• We can only read the specific files you pick in the Google Picker — we cannot access anything else in your Drive
• Excel / Word files are temporarily converted to Google format for reading and immediately deleted afterwards — your original file is untouched
• Authorization lasts 180 days; you'll need to reconnect after long inactivity, revocation, or password change

Content limits

• Connections per AI (Sheet + Doc combined): 2 on Pro / 5 on Business
• Google Doc / .docx character limit per file: 2,000 (content beyond is truncated)
• Google Sheet / .xlsx row limit per file: 300 rows on Pro / 600 rows on Business (rows beyond the limit are not read). Note: the bigger the sheet, the more content is sent to the AI on every turn; on a very large sheet, a turn that misses the prompt cache may cost extra credits

How to connect

1. On the AI's Knowledge page, click "Connect a cloud file" under "Teach it something new"
2. Click "Connect Google Account" (first-time use requires authorization)
3. Pick the file(s) yourself in the Google Picker; if a spreadsheet has several tabs, the preview card shows a "How to read" row: first tab, last tab (handy for one-tab-per-month schedules, a new tab is picked up automatically), or a specific tab. After connecting, you can change this any time by clicking the cloud file's card on the Knowledge page to open its preview
4. The system shows a preview; click "Add to knowledge base" to finish

Chat Guide

Tips and advanced features for chatting with your AI.

Chat Interface

• Left sidebar：AI selector and conversation history
• Main area：Message display
• Input box：Type your question, press Enter or click Send

Asking Tips

Better questions lead to more accurate answers from the AI:

Chat Operations

References

When the AI answers a question, a "References" section may appear below the answer. This tells you which document and paragraph the AI found its answer from.

References are important because they let you:

• Verify whether the AI's answer is correct
• Understand where the answer comes from
• Identify content in your files that needs updating

Building a High-Quality AI

Documents uploaded, character configured, but the answers still feel "almost right"? This guide collects the practical know-how that takes an AI from "working" to "great": how to prepare your knowledge base, tune the character, and verify quality before and after going live.

Core Concept: Knowledge Sets the Ceiling, Character Sets the Style

Answer quality comes down to two things: the knowledge base determines whether answers are correct and complete, while the character determines whether they sound the way you want. Both matter, but in this order — the knowledge base is the foundation, the character is the interior design. No decoration can fix a shaky foundation.

Four Rules for Your Knowledge Base

1. Small and focused beats big and messy — Only include what this AI needs, and remove outdated or unrelated files without hesitation. The leaner the knowledge base, the more completely the AI masters it, and the more stable the answers.
2. One AI, one topic — Need both customer support and internal training? Create two AIs, each with its own knowledge. Mixed topics make the AI confuse them and distract visitors with irrelevant answers.
3. One version of the truth — If old and new price lists coexist, or two files state different rules, the AI may cite the wrong one. When content changes, delete the old file first, then upload the new version.
4. Write concrete answers, not vague ones — With "shipping depends on the region", the AI can only repeat that sentence. With "NT$60 domestic, NT$120 for outlying islands, free over NT$1,000", it can actually answer visitors' questions. Numbers, dates, conditions, exceptions — the more specific, the better.

Document Writing Tips

1. Use headings; one paragraph, one idea — The AI understands content paragraph by paragraph. Clear headings and paragraph breaks help it locate the right information precisely.
2. Write FAQs directly in Q&A format — Write questions the way visitors would ask them. Q&A is the highest-hit-rate format — see the example in the "Document Management" chapter.
3. Upload tables as Excel / CSV — Structured data like price lists and spec sheets is cited far more accurately from a spreadsheet than buried in article paragraphs.
4. Avoid image-only scans — The AI reads text. A scanned PDF made entirely of images is unreadable on the Free plan (the Pro and Business plans include image text recognition).

Two Settings That Sharpen Your AI

The Knowledge page has two easily overlooked settings with a big impact on quality:

1. Review the "AI's Expertise" keywords — After you upload content, the system auto-generates this keyword set. It defines what the AI considers "on topic" — whether to steer a drifting conversation back and when to politely decline unrelated questions. Review it and refine it into 5-6 precise keywords: too broad, and the AI chats about everything; too narrow, and it declines too often.
2. Mark your most critical files as Important — Each AI can mark up to 3 documents as important, and it will prioritize them when answering related questions. The slots are precious — reserve them for what matters most, like price lists, flagship product intros, or current promotions.

Five Keys to Character Setup

1. Align the character with the knowledge base — A knowledge base about tea paired with a barista character leaves the AI torn. The character's expertise and audience should match the knowledge content.
2. Make the tone concrete — "Friendly and professional" is too abstract to work. "Like a neighborly big sister, uses everyday metaphors, avoids jargon" — that's a spec the AI can actually follow.
3. Fill in at least one Cognitive Layer field — Core beliefs, formative stories, internal tensions — these three fields upgrade the AI from "talks like a person" to "thinks like a person". Even filling in core beliefs alone makes answers noticeably more dimensional.
4. Set behavioral limits up front — Proactively define forbidden topics and capability boundaries. For example, a support AI shouldn't discuss politics or promise compensation on the company's behalf; a wellness advisor shouldn't make medical diagnoses.
5. Pick the right Answer Scope mode — For support and professional consulting, use Focused Mode — when wrong answers are costly, it's better for the AI to politely decline. For role-play and general assistants, Flexible Mode makes the conversation feel more natural.

Before Launch: Be Your Own First Visitor

Before sending out the share link, run a full round in Chat Test. These five checks catch most quality issues:

After Launch: Great AIs Are Grown, Not Built

A high-quality AI isn't a one-time setup — it's tuned continuously based on real conversations. Build this loop:

1. Check "What people ask" in Chat Logs in the left sidebar (Pro and above) — What visitors actually ask often differs from what you expected. Top questions are your knowledge base's to-do list.
2. Review "Questions it didn't answer well" and thumbs-down conversations — Every poorly answered conversation points to a knowledge gap or a blind spot in the character setup.
3. Add the missing answers to the knowledge base — Q&A format is the fastest patch — use the visitor's original phrasing as the question and the correct information as the answer.
4. Clean out stale content regularly — Prices changed, promotions ended, specs updated — sync the knowledge base and delete the old versions.

Share Pages

Create share pages so anyone can use your AI without signing up. Supports Chat and Smart Presentation page types.

Page Types

• Chat Page — Visitors chat directly with your AI, ideal for support, FAQ, and consulting
• Smart Presentation — AI transforms PDF into interactive presentations, ideal for reports and training (Pro plan)

Create Share Page

Access Control

Who Can Enter & the Guest List

On the Business plan you can make a share page require an email before use. Visitors enter their email and the code sent to it, and they appear on that page's guest list automatically. You never type an email by hand. Each share page has its own list.

On the share page settings page (the gear on that share page's row in the AI Overview), the "Who can get in" section has an "Email verification" row with a switch. Off means public: anyone with the link can use it. Turn it on and the access modes and limits appear (click "Adjust approval" to expand them).

Three access modes

Auto-approve and I'll review can also be limited to emails ending in a given domain, such as @company.com. Handy for internal training material, handovers, or campus resources.

Two layers of limits; whichever hits first stops access

• Whole page: seats, total query limit, expiry date. When reached, everyone stops.
• Per person: queries per person, days valid after approval. When reached, only that person stops.

Each limit has its own switch; off means unlimited. When both seats and per-person queries are set, the panel shows whether they exceed the page total. There is also "how many people can leave an email per hour": off means the default of 500, which stops someone from using your page to blast verification emails; turn it on and raise it before a big event.

What the guest list page does

• Entry points: "View list →" on the "Email verification" row of the share page settings page, "See the list →" under that share page's row in the AI Overview (it shows how many are let in and how many are waiting), or the "{n} waiting for your OK" alert at the top of the AI Overview.
• People waiting for approval are grouped in the "Waiting for you" band at the top, each with Approve and Reject, plus "Approve all". Requests older than 7 days expire automatically and the person can apply again.
• Click a person to open their details and set them individually: add queries, extend days, change the expiry date, reset usage, let them re-verify on another device, or suspend. Overridden people are marked and can be reverted to the page defaults anytime.
• Click "Select", then click people on the list to add queries, extend, approve, or suspend in bulk. "Paste roster" in the top-right ("Check roster" in I'll review mode): copy a column from Excel or Google Sheets, or pick an Excel/CSV file; names and phone numbers are fine, only emails are picked up. You preview first, then confirm: anyone waiting for approval who is on the roster gets approved, and the rest can be rejected in one go. The "⋯" menu at the top right has Copy share link and Export, and "Who can enter" takes you back to the share page settings.

What visitors see

Enter email → receive a code → type it in. Auto-approve starts right away; I'll review shows "Request sent, we'll email you once approved". All three modes work the same way: only approved people see the page content. Pending, suspended, expired, or out-of-queries visitors see the locked screen just like unverified ones, with the reason and who to contact. If seats are full, they're told before entering an email.

Web Embed

Pro plan and above can embed the AI into your own website. You bind your website domain first, and only bound websites can load it. There are two embed methods:

• iframe — Embed the full chat interface, ideal for standalone FAQ pages
• Widget — Show a floating chat button in the bottom-right corner, click to expand

Smart Presentation Mode

Pro plan and above offers Smart Presentation Mode. When you upload a PDF file, visitors see a dual-pane interface: the left side shows the PDF content, and the right side has the AI chat window. Visitors can read the document and ask questions side by side.

Customizing Your Share Page’s Look

You decide how each share page looks: apply a ready-made template, change colors and background, adjust the chat box style, or use the canvas-style drag-and-drop editor to freely arrange the layout, add decorations, and set up the mobile version separately. Even with no design skills, you can put together a polished page in 3 minutes.

Share Page Quota

Share page limits depend on your subscription plan:

• Free: 1 page
• Pro: 5 pages
• Business: Unlimited

Visitor Experience Flow

When a visitor opens a share page, they pass through the access gates you've configured. Steps you haven't enabled are skipped automatically.

Customize Share Page Appearance

You decide exactly how your share page looks. ShareYourAI offers two ways to edit — both on every plan: with no design background, Quick Templates gets you a polished page in 3 minutes; to fine-tune everything, the canvas-style Advanced Editor lets you drag, arrange, and design freely. Both edit the same settings, so you can switch from simple to advanced anytime and keep going.

Quick Templates (Simple Appearance Settings)

Open the "AI Appearance" popup with the "Quick design" button on the "What your share page looks like" card on the AI Overview page (the card appears once you have a share page; while setup steps remain, "Give it your look" at the top opens it too). No dragging needed, just pick and choose; it's the fastest way to go live. The popup has 4 tabs on the left:

• Templates — 12 built-in color themes (Nordic minimalist, Japanese zen, cyberpunk, Morandi, and more); one click applies a full coordinated palette. Only colors and visual style change — your layout stays put
• Brand — upload your AI avatar (synced to all share pages) and set a brand color (applied to buttons, message bubbles, and accents)
• Page — page background: solid color, gradient (with adjustable angle and stops), or an uploaded image (with fill mode and position controls)
• AI Style — describe the feel you want in one sentence under "Style description" (e.g. "warm sunset tones"), click "Generate style", and AI picks a full coordinated palette; there are ready-made style chips too

Good for: getting a new AI live quickly without fine-tuning every element

Advanced Editor (Canvas-Style Drag & Drop)

Switch in from the "Advanced editor" button inside the Quick Templates popup, or click "Advanced editor" on the "What your share page looks like" card on the AI Overview page. A canvas-based, drag-and-drop, what-you-see-is-what-you-get interface with 7 element types to design freely.

• Text — 18 fonts (including Noto Sans/Serif CJK, pixel, serif, and monospace faces), free font sizing, bold/italic, text shadow and outline; easily make headings and highlights
• Shape blocks — 26 shapes (rectangle, circle, heart, cloud, burst sticker, crescent, speech bubble…), filled with a solid color, linear/radial gradient, or a background image
• Images, buttons, dividers, plus YouTube video and Google Maps embeds; images can be flipped, rounded, and given borders and shadows
• Built-in stock libraries — cut-out stickers and decorations, a free commercial-use photo library, and ready-made backgrounds; just drag one onto the canvas
• Color tools and layers — gradients, opacity, eyedropper color picking; reorder, lock, and hide elements like pro design software
• AI Colors — describe what you want in a sentence and the AI sets up a full color scheme and visual style (background, message bubbles, chat-area styling); you place the images and decorative elements yourself

Good for: building a distinctive brand, executing your own design vision, adding decorations or text emphasis

Advanced editor interface tour

The advanced editor has four areas: the left toolbar (9 panels), the top toolbar, the central canvas, and the property drawer on the right. The left toolbar, top to bottom:

Top toolbar, left to right: back, undo/redo, the "Layers" panel, and canvas zoom; in the middle, the "Mobile Auto" toggle and desktop/tablet/mobile viewport switch; on the right, "Preview" (opens the visitor view in a new tab), "Save", and "Publish".

When you select an element, a floating toolbar appears above it with the most-used settings (colors, font size, shapes, etc.). Finer settings live in the property drawer on the right — open it by double-clicking the element or clicking "More properties" on the floating toolbar.

Canvas basics

The canvas is intuitive to pick up. Here are the most common actions:

Visual Details You Can Fine-Tune

Beyond the overall color scheme, these chat-related details can each be adjusted individually so the conversation view fits your brand:

• Background — two types, color block or image (switch between solid and gradient inside the color picker); a background image can take a semi-transparent overlay for clearer text, plus a focal-point adjustment so the subject isn’t cropped on mobile
• Message bubbles — set color, font size, and corner radius separately for visitor and AI bubbles; also adjust the conversation alignment (same side or both sides), max bubble width, and message spacing
• Input box — adjust the color and corner radius of the input field and send button
• Greeting — set the bubble text the AI uses to greet visitors the moment they land
• Signature — show “Shared by: your name” so visitors know who built the AI

Apply a Ready-Made Industry Template in One Click

No design skills needed. The system ships with 50+ full-page templates — and counting — covering scenarios like food & beverage, courses, fitness, e-commerce, arts, corporate branding, and event announcements. Pick one you like, apply it, then swap in your own text and images.

Three Ways to Present the Chat Box

You decide how the chat box appears to visitors — pick one based on the page’s purpose:

• Full block — the chat box is embedded right on the page with adjustable position and size (default). Best for chat-first pages where visitors should start asking immediately
• Bottom-right bubble — the chat collapses into the corner and expands when tapped. The whole layout is freed up for your intro, product shots, or campaign details — like a typical website’s support button
• Floating input — only a single input bar sits at the bottom; the chat panel slides in from the side after a visitor sends a question. Great for a one-page landing page where Q&A is secondary

AI colors

The "AI Assistant" panel on the left lets AI build a complete color scheme for you:

• Click "Open AI color generator", describe the vibe you want in at least 5 characters (e.g. "dark tech feel"), and choose how to handle the background: Keep (leave it as is) or Recolor (recolor it along with everything else). To change the background image or add decorations, use the stock photos, design assets and decoration panels on the left, or upload your own images
• AI colors only adjusts colors and visual style — it never moves your layout. Not happy with the result? Just undo. Generation has a daily limit that resets the next day; a run that started but didn't produce colors may still count once

Turn Elements Into Clickable Interactions

Images, text, buttons, and color blocks on the canvas can each be set to do something when clicked, turning the share page into an interactive landing page:

• Open link — go to your website, cart, sign-up form, or any URL; choose to open in a new tab or the same page
• Instant contact — one tap to add you on LINE, call, or email (with a preset subject). A LINE @ID, plain ID, or line.me link opens straight away; any other domain shows visitors a leaving-site notice first (when the AI is embedded on your own site, links back to that same site skip the notice)
• Share page — one tap opens the share menu so visitors pick where to send this page; unsupported devices copy the link instead
• Ask the AI — clicking auto-fills a preset question into the chat box (e.g., “I’d like to know about the course”) to prompt visitors to ask

You can also embed YouTube videos and Google Maps directly — great for product intro videos or a physical store location.

Tune Desktop, Tablet, and Mobile Separately

The same share page can have a different layout on desktop, tablet, and mobile. Switch between the three device modes at the top of the editor to fine-tune each:

• Mobile Auto — the toggle in the middle of the toolbar (on by default): while it's on, mobile and tablet visitors automatically get a full-screen chat interface with no layout work needed. Turn it off to customize small-screen layouts yourself
• Switch device mode — after switching to mobile or tablet, adjust each element’s position and size on small screens without affecting desktop
• Hide elements — some decorations get too cramped on phones, so you can set “Hide on mobile” or “Hide on tablet” to show them only on larger screens
• Automatic mobile layout — your free-form desktop layout converts to a top-to-bottom flow on phones; you can also let mobile visitors see your intro page first and tap to slide open a full-screen chat
• Arrangement — in the vertical flow layout on mobile/tablet, dragging an element freely pins it where you drop it; to restore auto-arrangement, switch "Arrangement" back to auto in that element's responsive settings

Saving & publishing

Edits never go live directly — they follow a draft → publish flow:

Common how-to questions

API Integration

Build AIs with your own AI tool (MCP)

On the Pro plan or higher, your own AI tool (Claude, Codex, Cursor or any tool that supports MCP) can connect to ShareYourAI and build a complete AI for you through conversation: add knowledge, set the persona and greeting, design the share page, create skills, and create and adjust share links (including short URLs and presentation mode). No code and no API key needed.

Create API Key

API Key Permissions

API keys currently have a single permission, applied automatically when a key is created:

Authentication

All API requests require an API Key in the Header:

Authorization: Bearer YOUR_API_KEY

Chat Query

The API supports both streaming and non-streaming modes.

To attach images: add an images array to the /api/v1/chat body (data URLs or base64, up to 5 images of 5 MB each; JPEG, PNG, WebP, GIF). The OpenAI-compatible endpoint takes standard image_url content parts, data URLs only. Sending more than 5 images, or an image over 5 MB, returns an error right away (no conversation is charged) — resize and resend. Images in an unsupported format are skipped without breaking the conversation.

Non-Streaming Mode

Waits for the complete AI response and returns it as a single JSON object. Ideal for simple integrations.

curl -X POST https://shareyourai.ai/api/v1/chat \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Hello, what is the return policy?"
  }'

Response Format

{
  "response": "AI 的回答內容...",
  "conversation_id": "uuid-string"
}

Streaming Mode (Recommended)

Set stream: true to receive response chunks in real-time via SSE (Server-Sent Events).

curl -N -X POST https://shareyourai.ai/api/v1/chat \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Hello, what is the return policy?",
    "stream": true
  }'

const response = await fetch('https://shareyourai.ai/api/v1/chat', {
  method: 'POST',
  headers: {
    'Authorization': 'Bearer YOUR_API_KEY',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    query: 'Hello, what is the return policy?',
    stream: true
  })
});

const reader = response.body.getReader();
const decoder = new TextDecoder();

while (true) {
  const { done, value } = await reader.read();
  if (done) break;
  console.log(decoder.decode(value));
}

SSE Event Types

In streaming mode, the following events are sent in sequence:

Contacts, bookings and link cards over the API

The built-in skills you turn on in AI Skills (visitor contact capture, bookings, link cards) is available through the API too. Your system decides how it looks; the platform handles the data and the follow-through (notifying the owner, booking confirmation, booking code lookup). In streaming mode the fields arrive in the done event; in non-streaming mode they are part of the response JSON. The fields are the same:

Integration flow (four steps)

1. Call /api/v1/chat as usual and always pass the conversation_id from the previous response on later turns. Contact capture, bookings and booking-code lookup all rely on it to recognise the same conversation.
2. When the visitor has given all the details, the done event (or the non-streaming response) contains lead_pending. Show its collected fields and summary to the user with a "Looks right" button; if it is absent there is nothing to do.
3. When the user confirms, send the lead_pending content together with conversation_id to POST /api/v1/chat/lead/confirm unchanged. Do not rebuild collected yourself — keep the field names exactly as returned.
4. Show the result based on the returned flags: only notified=true means it was delivered. Display visitor_code to the user — quoting it later in the chat looks up or updates the booking. Nothing else is needed on later turns; the AI knows it has been sent.

Example done event (details complete)

{
  "conversation_id": "conv-12345",
  "lead_pending": {
    "intent": "booking",
    "collected": { "name": "王小明", "phone": "0912345678" },
    "summary": "想預約週六下午看屋",
    "reserve": { "date": "2026-09-12", "time": "15:00" },
    "cancel": false
  },
  "action_card": { "type": "map", "title": "地點導航", "url": "https://maps.app.goo.gl/..." }
}

Submit contact / booking details

Once the user confirms, send the lead_pending content together with conversation_id to this endpoint — only then is the AI owner notified. Re-sending within the same conversation within 24 hours counts as an update to the same lead. When notified is false, check dedup / quota_exceeded / notify_capped and show the matching message instead of a success state.

curl -X POST https://shareyourai.ai/api/v1/chat/lead/confirm \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "conversation_id": "conv-12345",
    "intent": "booking",
    "collected": {"name": "王小明", "phone": "0912345678", "preferred_time": "週六下午"},
    "summary": "想預約週六下午看屋",
    "reserve": {"date": "2026-09-12", "time": "14:00", "adults": "2"}
  }'

Example confirm response

{ "notified": true, "updated": false, "dedup": false, "quota_exceeded": false,
  "notify_capped": false, "capped": false, "loop": true, "visitor_code": "SY-K7M2-4QX" }

Confirm response flags

Notes

• action_card and item_cards are just link data — render them as buttons or cards; what happens on click is up to you. item_cards.image_url is a permanent image URL you can display directly; it only stops working once the card is deleted.
• lead_pending.cancel = true means the visitor wants to cancel an existing booking; pass it back to confirm unchanged and the platform runs the cancellation flow.
• If the owner has not turned on the built-in "Leave contact info" skill on the AI Skills page, confirm returns 403 — check the AI settings before integrating.
• The OpenAI-compatible endpoint (/v1/chat/completions) does not carry these fields yet; use /api/v1/chat when you need contacts, bookings or link cards.

OpenAI-Compatible Endpoint

Also available at /v1/chat/completions, fully compatible with the OpenAI API format. Works directly with n8n, Dify, Zapier, LangChain, and more:

curl -N -X POST https://shareyourai.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "default",
    "messages": [{"role": "user", "content": "Hello, what is the return policy?"}],
    "stream": true
  }'

Continuous Conversation

Pass the conversation_id from the previous response into the next request to maintain conversation context:

curl -X POST https://shareyourai.ai/api/v1/chat \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "那價格呢？",
    "conversation_id": "uuid-from-previous-response"
  }'

API Quota

API Key limits depend on your subscription plan:

• Free plan does not support API
• Pro: 5 keys
• Business: Unlimited

Security Tips

1. Never expose API Keys in frontend code — use them on the server side only
2. Use the principle of least privilege — only grant necessary permissions
3. Rotate keys regularly — set reasonable expiry times
4. Monitor usage — check API statistics regularly and disable keys if anomalies arise
5. Store securely — keep API Keys in environment variables, not in source code

Analytics

Track AI usage, analyze conversation trends and visitor behavior.

Key Metrics

At the top of the Analytics page (left sidebar), one big number shows how many questions were asked in the period, followed by:

What Analytics Can Do for You

• Understand usage patterns — See how many people use your AI daily and how many questions are asked
• Optimize answer quality — Find which questions the AI struggles with and improve your knowledge files
• Monitor API usage — Track API call counts and conversation credit consumption
• Plan capacity — Decide whether to upgrade based on usage trends

Usage Data

• "Questions per day" line chart — toggle the All, Visitors and API lines; switch Today / Last 7 Days / Last 30 Days / Last 3 Months at the top right, and pick all AIs or one from the "Showing" menu
• "Most-asked AIs" ranking — click an AI to see its numbers
• Your monthly points are always shown on the plan card at the bottom of the left sidebar; this page only shows a notice when they're running low or used up
• API usage monitoring (show only the "API" line on the Analytics chart)

Chat Logs

Chat Logs in the left sidebar has two tabs, "What people ask" and "Every conversation":

• In "Every conversation", filter by All / Thumbs-down, time range and source (share page / LINE / API), and pick an AI from the "Viewing" menu
• Open any conversation to read its full content, with source, IP, and time
• Share page, API and LINE conversations in one list grouped by date — click "Show more" to load more
• "What people ask" (Pro and above) sums up the last 30 days: questions it didn't answer well, the most-asked topics, and who is asking, so you know what to add; click "Refresh" to update

Subscription & Billing

Not Sure Which Plan?

Plan Comparison

For the latest monthly fees, yearly discounts, and pricing (in TWD), see the subscription page. The table below matches the plan cards there.

Upgrade & Downgrade

• Upgrade：Effective immediately after payment
• Downgrade：Effective at end of current billing period; excess resources paused, not deleted

What Happens After Downgrading?

After downgrading, your data won't be deleted. However, if your usage exceeds the new plan's limits, the excess will be "paused" — your data is still there, just temporarily inaccessible. You can restore access by upgrading again or removing some resources.

Quota Explanation

ShareYourAI has two types of usage limits:

• Monthly quota — Resets automatically each month, like conversation credits. Restored to your plan's limit every month.
• Total limit — Doesn't reset automatically, like AI count, file count, storage. You need to delete old items to add new ones.

Top Up Visitor Conversations

Used up your monthly visitor conversation quota but not ready to upgrade? You can buy additional conversation credits — once your monthly quota runs out, the system automatically draws from your top-up balance.

When does it make sense to top up?

• A few days left in the month and you're about to run out
• Running a campaign or event with a temporary traffic surge
• Short-term project — you won't need the extra capacity afterwards

Top-up Packs

Consumption & Expiration Rules

• Monthly quota first, then top-up: visitors burn through your plan's monthly quota first; top-up balance only kicks in once the monthly quota hits zero
• 12-month validity: each top-up is valid for 12 months from purchase; any unused balance expires after that
• Stackable: you can top up multiple times — all balances combine into one pool
• Everything is currently settled in New Taiwan Dollars (TWD): cards issued outside Taiwan work too, with the exchange rate and any foreign transaction fee set by your card issuer; USD pricing isn't available yet

Auto Top-up: refill automatically when credits run low

Don't want a mid-month "out of credits" interruption? On the Top-up page (account menu at the bottom left), find the Auto Top-up card, turn on the switch, pick your three settings, and hit Save. From then on, when your remaining credits drop below the threshold you set, the system buys a pack with the card bound to your subscription.

• You decide all three settings: when to refill (below 50 / 100 / 300 / 500 credits), which pack (Starter / Standard / Bulk), and how many times per period (1 / 3 / 10, or no limit)
• Every automatic charge emails a receipt and e-invoice; auto-purchased packs work exactly like manual ones — valid 12 months, used only after the monthly quota is exhausted
• If a charge fails (e.g. an expired card), Auto Top-up pauses immediately and emails you — it never retries on its own. Update your card and turn it back on on the Top-up page
• Once your per-period limit is reached, no more charges are made; when credits run out, the AI stops replying to visitors until you raise the limit, top up manually, or the monthly quota resets. You can turn it off anytime — purchased credits are unaffected

Notifications & History

• You get an email when your monthly quota is running low, suggesting a top-up
• You get another email when your top-up balance is running low
• You get a reminder 30 days before any top-up expires
• See when each pack expires under "When each pack expires" on the Top-up page; receipts and billing history are in User Settings › Payments

How to top up

1. Click "Buy points" on the plan card at the bottom of the left sidebar
2. Or open the Top-up page from the account menu at the bottom left and choose a pack under "Add a pack"
3. Pick a pack, confirm, and pay with the credit card already linked to your subscription (same card)
4. After payment, the system automatically sends a receipt and electronic invoice via email

Account & Security

Manage your profile and account security. All of the settings below live in "User Settings": open the account menu at the bottom of the left sidebar and choose "User Settings" (it has Profile, Account security, Payments, Notifications and Connected apps).

Profile

In "Profile" you can manage (a save bar appears after you change something — press Save to apply):

• Account email: cannot be changed after registration
• Nickname & avatar: nickname is freely editable; when signing in with Google, the avatar comes from your Google account
• Language preference: Traditional Chinese or English, applied across the interface
• Company info (optional): company name, tax ID, and VAT number, used for invoicing
• Brand URL (Pro and above): set a custom brand handle so your share pages get a memorable URL (e.g. shareyourai.ai/@yourbrand). After a downgrade your handle is reserved for 60 days — upgrade again within that window to keep it; afterwards it is released for others to claim

Two-Factor Authentication (2FA)

Two-factor authentication adds an extra layer of protection: even if someone gains access to your login, they can't get into your account without the code on your phone. Once enabled, every login requires the 6-digit code from your authenticator app.

How to Enable

1. Open the account menu at the bottom left → User Settings → Account security, then click "Turn on" next to "Two-step verification"
2. Install an authenticator app on your phone (Google Authenticator, Microsoft Authenticator, Authy, and others are supported)
3. Scan the on-screen QR code with your authenticator app (you can also enter the key manually if scanning isn't possible)
4. Enter the 6-digit code shown in the app to finish enabling

Backup Codes

After enabling 2FA, the system generates a set of 8-digit backup codes. When your phone isn't at hand or the authenticator is unavailable, you can log in with a backup code instead:

• Each backup code can only be used once — it expires after use
• Backup codes are shown in full only once — download and store them safely right away
• You can regenerate them at any time, but all old codes become invalid

Lost Your Authenticator?

If you switched phones or deleted the authenticator app:

• Still have backup codes: log in with a backup code, then use "Turn off two-step verification" in User Settings › Account security and set it up again
• No backup codes either: on the login verification page, click "Can't use your code?" to request a reset. The system enters a 72-hour cooldown and sends a notification email to your inbox — if the request wasn't made by you, you can cancel it from the email. After the cooldown ends, 2FA is automatically disabled and you can log in normally.

Notification Settings

In "Notifications" under User Settings you can choose which email notifications to receive (changes apply right away); emails about visitors leaving contacts, bookings or files are set under Notifications at the top of each AI's Skills page:

• Security login alerts: get notified when a new device logs into your account
• Subscription expiry reminders: get notified before your subscription expires
• System announcements & feature updates: system updates, maintenance, and new feature news

FAQ