Stop searching for the same answer.
Keep approved answers in one library so AI can use the right policy for a routine question.
AI SUPPORT FOR GAMING OPERATIONS
AI answers from your approved knowledge and rules. A new question may need your team once: review the answer, save it and test the same question or its approved wording variants. You control which replies need approval.
Try it with sample data. Live channels and account connections need setup and testing.
Sample record matched. AI answers the status question.
What to compare in your pilot: staff minutes per case, first-reply time and the share answered by AI.
Explore pricing and plan your scopePlanning estimate · Your assumptions, not measured client results
Change the three assumptions to compare the same daily chat volume.
Calculation: chats × staff minutes ÷ 60. After = before × (1 − AI share). Assumes the remaining chats take the same time.
Excludes setup, monitoring and exception follow-up time. An estimate, not guaranteed savings; measure the result in your pilot.
YANDEE SUPPORT, EXPLAINED
See AI answer routine questions, follow your approval rules and reuse a new answer after review, testing and publication. Illustrations and local demo screens use sample data.
Download the service explainerEnglish narration · English captions
Service explainer · Local demo and illustrated workflows
In the film, “Approval” means DRAFT review or a rule requiring a person. The actual modes are AUTO, DRAFT and OFF.
Music: “Effervescence” by Scott Buckley · Edited excerpt · Scott Buckley · CC BY 4.0
Another late-night queue. The same questions. Answers spread across tabs. Then the next shift starts all over again. Meet Yandee: AI that answers routine questions and checks records, following the rules you set. When the answer is already approved, AI replies. Your team does not need to type the same answer again. You choose the controls: answer automatically, prepare a draft, or ask for approval. Some sensitive cases always need a person. A new question? Your team answers once. Review, test and publish that answer. AI can then handle matching questions next time. Manage each brand in its own workspace, with its own team and knowledge. Start with your policies, review the answers, and plan a Telegram-first pilot. Test delivery before going live. Show us where your support gets stuck. Let’s map your first Yandee pilot.
Try approved AI replies, reviewed knowledge reuse and AUTO/DRAFT/OFF rules with sample data. Live hosting, Telegram, backend access and action-specific approval flows are configured and tested before launch.
SEE IT FOR YOURSELF
Try an approved answer or a sample status check. Then review a new question, save the answer as knowledge and test the next repeat. Your rules decide when AI replies and when approval is needed.
Keep approved answers in one library so AI can use the right policy for a routine question.
Open the interactive demo to see AI answer eligible questions. Staff review mode remains available when your policy requires it.
When review is required, the question and checks stay together. A new routine answer can then be reviewed, saved and tested for reuse.
Yandee interactive demo · Sample data
LOOK INSIDE YANDEE
Actual screenshots from the interactive Yandee demo. A clear queue, a focused conversation and customer context beside it. Explore the same screens yourself with sample data.
See waiting conversations, AI activity and the next case to pick up. Open the detailed report when you need it.
The customer stays at the centre. Queue on the left, customer details on the right, a reviewable draft below. Extra tools open when you need them.
Find the conversation, its assignment and its current state in the left panel.
Read the customer’s question and inspect the private reply before anyone sends it.
Check customer context and the latest read-only record on the right. Open investigation details in the conversation.
Explore the setup checklist in the same workspace. The demo keeps live channel and backend activation separate from completed sample steps.
Inspect AI modes and channel settings in one place. These controls simulate the policy you would configure for a verified pilot.
The interactive demo has English, Thai and Chinese controls. Screenshots below show sample data; detailed scenario and audit text may use another language.
LESS SEARCHING BETWEEN REPLIES
Before: staff search and type the same answers every shift. With Yandee: AI handles eligible routine questions. New cases feed a reviewed knowledge process, so a repeat does not always need another handover.
The customer has sent a slip. Your team still has to find the account, check the transaction and explain what is known.
In the matched-record demo, AI checks the sample account, amount and reference, then tells the customer the recorded status. Staff only investigate when the record is missing or does not match.
Sample records in this demo. Your real backend needs a tested read-only connection.
No current matching record. Keep the case open and ask the team to check.
Workflow illustration · Sample data
Support hours, account access and operating rules keep coming back. Searching old chats for each answer takes attention away from unresolved cases.
AI uses the approved guide to answer routine questions. If a question is new, your team checks the answer and saves it for review and testing. Once published, AI can use that answer for eligible repeats under your rules.
Try a routine answer in the local demo. Live sending is tested during setup.
Approved policy → AI answer
Permitted record → visible context
Case reason → the right person
Requirements → setup checklist
Workflow illustration · Sample data
One team may cover several brands. A reply copied from the wrong workspace can carry the wrong rules.
Organize each brand’s knowledge, channel and team access in its own workspace. A visible brand context helps reviewers check the answer against the correct policy.
Separate workspaces. Combined cross-brand inbox is planned.
CURRENT WORKSPACE
Team + knowledge + channel
Team + knowledge + channel
Team + knowledge + channel
Workflow illustration · Sample data
“Any update?” comes in again. A guessed arrival time may calm the chat briefly, then leave the next shift explaining a promise nobody verified.
In the demo, a verified sample record says the withdrawal is queued for review. AI explains that reason without inventing an arrival time. If the record is missing or unclear, staff take over.
Read-only status answers in the demo. AI does not approve withdrawals or change balances.
Workflow illustration · Sample data
The customer has explained the problem. The next person still needs to know what was checked, why it was handed over and who owns the next action.
Keep the conversation, available evidence and handover activity together. A person takes ownership, while account closure and sensitive concerns stay with trained staff.
Assistance supports trained staff; it does not replace their judgment.
“I want to close my account. Please help me.”
Account closure requires a staff decision. AI does not close the account.
Workflow illustration · Sample data
A new question reaches your team. If the answer stays buried in that chat, the next shift has to solve it again.
Save the team’s answer as a knowledge draft. Review and publish the right version, then test the same question and its approved wording variants. AI can reuse that answer for those matching questions when your rules allow it.
Reviewed knowledge reuse is available in the demo. Automatic learning and publication from every chat are not enabled.
Approved knowledge + your rules
Workflow illustration · Sample data
A CLEAR VIEW OF THE PRODUCT
Run a routine question or sample status lookup, then compare it with a case that needs staff.
English and Thai flows for account review, workspace setup and a clear next step.
A simulated offer, invoice and setup journey. No money is collected in the demo.
Telegram, backend access, hosting and payment providers are scoped and verified for your pilot.
THE SCREENS BEHIND THE SERVICE
Open work, human ownership and setup status give the next shift a practical starting point.
Waiting conversations and sample activity are separated from detailed reporting controls. All figures come from this synthetic demo, not customer performance results.
The same responsive demo, with navigation and customer context available when needed. No native app installation is required.
Desktop and mobile screenshots use synthetic records. The demo is a responsive website, not a native mobile app.
FOLLOW THE SUPPORT WORK
Choose a sample case, inspect its evidence, then see the exact reply or handover.
Choose a sample deposit outcome
I sent the transfer, but my balance has not updated. Can you check?
No matching deposit is confirmed in the available record. Please give the transfer reference and time through the approved support route. I’m handing the case to staff to verify the account and record; I cannot confirm that the funds arrived.
Illustrative scenarios with sample data. This page is not connected to a customer account or live backend.
Your knowledge, made useful
When a new question needs your team, save the checked answer with its source. Review and publish it, then test the repeat question. AI can reuse approved knowledge when the question, brand and rules match.
OWNERSHIP YOU CAN SEE
A case that needs approval keeps its question, source and next step together. Your team handles that decision; approved routine answers can stay with AI.
Captured from a synthetic conversation after staff takeover. No message was sent to a real customer.
CONTROL IS PART OF THE WORKFLOW
AUTO sends eligible replies. DRAFT prepares them for approval. OFF stops AI. The demo has workspace, channel and case controls; action-specific approval rules for live use are agreed and tested during setup.
Organize each brand’s team, channel and approved knowledge in its own workspace. Define the access and deployment needs for your pilot.
Scope permitted backend lookups with read-only access. No balance changes or withdrawal approvals.
READ-ONLY BY DESIGNChoose draft review where a person must approve the reply. Keep human handling on knowledge or cases that require it, and use AUTO for eligible routine answers.
YOUR APPROVAL RULESSee the sources, reviewer decisions and activity trail. Give the next shift a clear starting point.
Responsible-gambling concerns, account risk and uncertain payment cases belong with trained staff. The pilot is built around that handover.
THE COMPLETE WORKSPACE PICTURE
Open a category to see the workflow and its current scope. We agree on the exact capabilities for your pilot before rollout.
Use AUTO for eligible approved answers, DRAFT when a reply needs approval and OFF to stop AI. Workspace, channel and case settings control the local demo. Live action-specific approval rules require scoping and testing.
Save new answers and corrections as drafts. Review, publish and test them so AI can reuse the approved version for eligible matching questions. Update or revoke knowledge when the rules change.
A case keeps its context, owner and next action. Human handover and support-ticket closure are distinct from payment confirmation.
Time-bounded maintenance and incident notices are a future extension. Your pilot starts with reviewed operating policies and staff handling of exceptions.
Voice transcription and bank-slip verification are planned capabilities, not ready-to-buy add-ons. Each needs separate scoping and validation before live use. We do not infer a completed transaction from an attachment.
Inspect the answer and source, take over exceptions and follow the activity record. Choose staff review for workflows that require it.
Use the records and fields permitted for the case. Confirm identity and access before bringing account-specific information into a reply.
Knowledge assistance, read-only lookup, human handover and commercial onboarding have separate responsibilities. Backend lookups require a tested connector.
Telegram first. Website chat and other inbox channels are assessed per pilot. A channel is ready only after delivery and handover tests pass.
Scoped roles, explicit approvals and separate author/reviewer responsibilities in commercial onboarding. Access is checked for the action and workspace.
Test a new answer against repeat questions and review the result before relying on it. This is supervised knowledge reuse. Automated evaluation and self-publishing from chats remain unimplemented.
Responsive web screens adapt to phones. Native iOS/Android apps, biometric unlock and mobile push are not part of the current pilot.
A workspace for each brand, with its own knowledge and permitted team access. A combined cross-brand customer inbox is a later capability.
Use authorized content, minimize personal data and keep credentials out of chat. Live hosting, retention and production identity requirements are agreed during setup.
“Demo” means an inspectable sample flow, not a connected production service.
FROM YOUR POLICIES TO YOUR FIRST PILOT
Bring the policies your team uses, choose the first cases and review the replies together. Test delivery and handover before the pilot reaches real customers.
Plan my support pilotScope and pricing are agreed around your requirements.
Choose the first support questions, team owner, brand and backend to assess.
Plan one Telegram connection. Test inbound messages, staff replies and handover before activation.
Choose which routine questions AI may answer, which replies need approval and which actions are outside its scope. Save reviewed answers as the starting knowledge.
Try approved replies, first-time questions and approval-required cases. Review a new answer, publish it and test the repeat before enabling eligible automatic replies.
Review actual cases together. Add a validated backend lookup or another workflow when the first one is ready.
FROM ENQUIRY TO OPERATION
BEFORE WE START
You can try the local demo now with sample data. To serve real customers, we first set up your channel, approved answers, account permissions and handover route, then test the complete workflow.
Telegram is the first channel in the pilot plan. For website support, we assess whether an inbox integration fits your setup. We confirm the connection and test real delivery before enabling your channel.
We first review the APIs and permissions your platform provides. Supported lookups can use a scoped, read-only connection. We do not promise compatibility with every backend, and the pilot does not change balances or approve withdrawals.
Yes. Authorized site content, policy documents and selected chat history, with personal data minimized, can enter the review process. Your team approves the material before publication. We do not treat every past answer as correct, and conversations do not become knowledge without review.
Yes, the local demo has AUTO for eligible replies, DRAFT for approval before sending and OFF. A new answer can be saved as a knowledge draft, reviewed, published and tested for future matching questions. This is reviewed knowledge reuse; chats do not automatically approve or publish themselves. Live connections and action-specific approval rules still need setup and testing.
We scope a workspace for each brand’s team, channel and knowledge, then confirm deployment and access requirements. Pricing is quoted for the agreed pilot, integrations and operating needs. No fixed savings or performance outcome is promised.
LESS REPETITIVE WORK STARTS HERE
Show us your repeat questions, busiest shifts and support costs. We’ll scope a pilot around the work you want to reduce, then measure what changes with your team.
Plan my Yandee pilothello@yandee.aiPlease don’t email passwords, access tokens or customer records. We’ll agree on a secure setup process.