DigiClient AI reads the orders you already have, finds the next sale worth making, and writes the message and the payment link for you. You read it and tap approve. The same grounded desk works across WhatsApp, LINE, website chat, and Instagram DMs where connected; native rich content follows each channel's rules.
Start your free trial 30-day free trial · No credit card required · No setup fee · Export your data any time
A real AI answers. Payment runs in Stripe test mode (you are never charged). Please do not enter real personal information — demo conversations are deleted automatically every day.
This is one of 11 demos →THE BUSINESS JOBS
Start with the change you want to see. Every lane opens a grounded replay, then shows the operator proof and next move.
Answer the question on your site or in chat, then carry the customer to a booking or a well-prepared human conversation.
Proof: Source-backed answers, booking guidance, and contextual handover
Channel: WhatsApp · LINE · web chat
Move from product choice to basket, verified details, exact-amount checkout, and one reliable sales record. The public replay uses Stripe test mode; tenant-owned payment adapters sit behind credential and readiness checks.
Proof: Basket read-back, masked detail confirmation, payment-provider readiness, checkout, and ledger record
Channel: WhatsApp · LINE · web chat · counter
Use purchase and response history to prepare a timely follow-up, then check consent, quiet hours, and frequency before approval.
Proof: Follow-up drafts, campaign expiry, approval, and attributed outcomes
Channel: WhatsApp · LINE (with consent and send-readiness checks)
Turn a photo, voice note, or receipt into a useful next action, and send a problem to a person instead of forcing a sale.
Proof: Catalog matching, transcription, receipt draft, and issue handover
Channel: WhatsApp · LINE (with a text fallback)
Enable only the capabilities you need, with a path back, while staff see the brief, ownership, approval, and outcome in one flow.
Proof: Guided setup, capability history, shared inbox, assignment, approval, and Stats
Channel: WhatsApp · LINE · web chat · phone
Open published information through a read-only storefront, then return booking and payment to the chat you control.
Proof: Off-by-default MCP storefront, public scope, no customer data, and chat handoff
Channel: AI agent → WhatsApp · LINE · web chat
The live phone above is one of 11 grounded business journeys. Each replay starts with a goal, shows the operator decision, and makes the product proof visible before you choose a live test, signup, or consultation.
THE PLATFORM TODAY
Choose a business job, then open the replay that shows the operator move and the expected impact. Live channel setup, consent, permissions, and provider readiness remain explicit where they matter.
Use your pages, products, campaigns, and verified order history. When a private record is not verified, ask for another identifier instead of guessing.
Proof: Source citations, private-record checks, and a safe handover when the source is not enough
See the grounded-answer demo →Capture a lead in the conversation, guide availability and booking, or give staff the context and SLA they need to take over. An explicitly enabled public mode can also answer first-time visitors within tenant-wide and per-user caps.
Proof: Lead capture, availability, shared queue, ownership, SLA-aware handover, and capped public replies
See the booking and handover demo →Recommend the right product, read back the basket, verify masked details, and take the customer to an exact-amount checkout in the conversation. The public replay uses Stripe test mode; tenant-owned payment adapters can be enabled after credential and readiness checks, with the paid state returned to the ledger.
Proof: Catalog matching, basket integrity, Stripe test checkout, payment-provider readiness, and payment-state confirmation
See the chat-checkout demo →Connect Shopify, chat orders, and in-store/POS sales so staff spend less time reconciling and can measure the outcome of each action.
Proof: Draft import and publish review, one sales ledger, and outcome Stats
See the unified-ledger demo →Find the next opportunity from purchase, visit, response, and member context. Consent, quiet hours, and frequency are checked before approval; stale offers and duplicate coupon use are blocked.
Proof: Follow-up drafts, campaign expiry, consent/frequency gates, member context, and coupon records
See the repeat-visit demo →Turn photos, receipts, and voice notes into a useful next action. When confidence is low, ask or hand over instead of inventing an answer.
Proof: Catalog matching, receipt draft, voice transcription, and issue handover
See the multimodal demo →Use guided setup, capability-level enable/disable and history, then keep the AI brief, ownership, approval, close summary, and outcome in one operating loop.
Proof: Capability state, change history, shared inbox, assignment, approval, close summary, and Stats
See the operator-console demo →Bring in website knowledge, use the standard AI or your own endpoint, and review product or marketing creative before anything is published.
Proof: URL ingestion, source review, BYOAI connection, and reviewed creative publish gates
See the AI-features demo →Instagram DMs use the grounded answer and human-handover flow. The public simulator sends nothing; live use needs a professional account, Meta permissions, webhook setup, and the 24-hour window. Rich content falls back to text.
Proof: Instagram-style DM replay, shared answer engine, handover preview, and an explicit live-readiness boundary
Try the safe Instagram DM simulator →Goal：Capture more booking intent
Impact：Move from availability to a confirmed booking in chat, while staff focus on the requests that need them.
Watch the replay → ConsultationsGoal：Keep high-value consultations moving
Impact：Let routine questions resolve quickly and give specialists the context they need to turn a consultation into an appointment.
Watch the replay → OrdersGoal：Shorten the path from interest to payment
Impact：Collect what a first-time buyer needs in the conversation and let a returning buyer confirm saved details before checkout.
Watch the replay → Unified ledgerGoal：Give every sale one reliable record
Impact：Bring Shopify, chat, and counter sales together so staff spend less time reconciling records and more time serving customers.
Watch the replay → RetentionGoal：Turn a maybe into a next opportunity
Impact：Keep the reason someone did not buy today, then approve a timely follow-up instead of letting the conversation disappear.
Watch the replay → Campaign resultsGoal：Make every promotion answerable
Impact：Keep the live offer consistent in replies and broadcasts, stop stale promises after expiry, and see which results came from it.
Watch the replay → ReadinessGoal：Start with only the capabilities you need
Impact：Turn on website knowledge, creative tools, or your own AI step by step, with a versioned path back when a setting changes.
Watch the replay → LeadsGoal：Stop losing new enquiries on your site
Impact：Put a grounded answer and a human consultation path beside the page a visitor is already reading.
Watch the replay → Multimodal serviceGoal：Remove friction from the way customers ask
Impact：Turn a photo, voice note, or receipt into the next useful action, while sending complaints to a person instead of upselling them.
Watch the replay → DiscoveryGoal：Create a safe new discovery path
Impact：Let agents find published information about your business while booking and payment return to the chat you control.
Watch the replay → Operator productivityGoal：Give staff one clear next action
Impact：Bring the conversation, AI brief, handover, approval, and outcome into one flow so fewer requests wait without an owner.
Watch the replay →Open the full demo library → or talk through the right starting point
A chatbot waits to be asked. DigiClient AI goes looking — through what your customers bought, when they bought it, and what they never came back for.
Orders, baskets, visits, the products someone buys every eight weeks and has not bought in twelve. Your data, in your account — not a shared model, and not a guess about your industry.
Not "consider reaching out". The real text, in your voice, with the right product, the right price, and a payment link that already works. Ready to send.
You read it and tap. Nothing goes out on its own — and every send is recorded against the order it came from, so you can see what the message actually earned.
WhatsApp and LINE remain the core owned-channel lanes. Website chat is your owned front door, and Instagram DM adds another customer entry point with a clear live-readiness boundary.
One subscription runs every number and account you have — including a separate staging account at no extra cost. Tested settings promote to production in one tap, and test conversations never mix into your real inbox. We do not bill per account.
What actually arrives in a chat is not tidy prose. It is a photo of the product, a receipt, a voice note recorded while walking.
A customer sends a picture of what they bought. The AI matches it against your catalog and sends back a card they can re-order with one tap. Same on WhatsApp and LINE. When it cannot find a confident match it asks which item you meant, instead of inventing one.
A voice message is transcribed and then joins the ordinary flow — bookings, product questions, FAQs, all handled exactly as if it had been typed. If the audio is unclear it politely asks for text rather than guessing.
An AI that answers a photo of a damaged item with "would you like another?" loses the customer. DigiClient AI reads the intent behind the picture and replies with an apology and a Talk to a person button. The customer just taps.
Customers increasingly ask ChatGPT and Claude where to book and what to buy. Every business on DigiClient AI can open its own MCP (Model Context Protocol) storefront so those agents can find it and recommend it — and it stays off until you turn it on.
Flip one switch and your shop gets a shareable MCP endpoint that any agent can connect to — ChatGPT connectors, Claude, or any MCP client. Agents browse your published catalog, your public pages, and your prices and openings, then hand the customer to your chat to book.
The agent door shares only what you have already published: catalog, public pages, rates. No customers, no conversations, no orders. Agents cannot book, buy, or change anything — the transaction always closes in your own chat.
Sharing is off by default. Turning it on goes through the same publish review as any other change; turning it off takes effect immediately. See it running today: the TickTock Town demo town is open to agents at ticktock-town.com.
Move the sliders. The arithmetic is not complicated — that is rather the point.
Conversion rates are yours to prove, and the exchange rates here are illustrative — this is an estimate, not a quote.
Not showing dollars first is deliberate: Meta has paused WhatsApp marketing templates to US and Canada numbers since April 2025, so the win-back motion this calculator prices does not run there. Utility and customer-service replies to +1 numbers, and web chat everywhere, are unaffected.
Everything below is a mechanism, not a promise in a footer.
WhatsApp campaigns can only go out on templates Meta has approved. There is no path in the product that sends an unapproved broadcast, so the rule cannot be forgotten under deadline.
Customer-service replies are free-form inside WhatsApp's 24-hour window; outside it, the product uses a template. It tracks which window a conversation is in so you are not guessing about a rule Meta enforces on your number.
Your ledger is your account's. Bring your own AI keys if you want to — the platform runs on them, and nothing about your catalog or your customers trains a shared model.
Answers are grounded in your own pages and products, and cite the sentence they came from. When the source does not cover a question, the AI says so and offers a human, rather than inventing a plausible price.
A test account, included. Test conversations never appear in your real inbox, and a production campaign cannot reach the contacts you invented while testing.
Every outbound campaign waits for a human tap. The AI drafts; you decide. The approval is recorded next to what it earned.
And seven guardrails decide what actually leaves the building — before it sends, when it sends, how often, and what is on the record afterwards.
Promotional sends are held automatically between 21:00 and 08:00 in the shop’s own timezone, and delivered once the morning opens. It is the default, not a setting somebody has to remember.
Marketing consent is recorded and managed in a ledger. Switch the consent gate on and a contact whose opt-in cannot be verified is not sent to — the failure direction is “don’t send”, not “send anyway”.
Cap how much one customer can receive, by rank. Public replies to people who never messaged you carry their own ceilings — per user and per rolling 24 hours — from the first day, without configuration.
Per-rank discount limits are the shop’s decision. AI replies, drafts and broadcasts alike: a promise above the ceiling is stopped by the mechanism. Margin is protected by a rule, not by the model’s good intentions.
Who was sent what, and when, is in the delivery ledger. Payment links keep their amount, recipient and settlement history per transaction. “Why was this sent?” has an answer you can read months later.
When something stops a send, the reason appears in plain language with the next step attached. Nothing fails silently and waits to be discovered in a report.
Your ledger exports on demand, and deletion requests are honoured without delay. Cancelling does not erase your data — it leaves with you, which is what makes leaving a real option.
No. They message you on WhatsApp exactly as they already do, or use the chat on your website. There is no app, no portal and no account for them to create.
No — the setup flow walks you through connecting one, including the Meta side. If you already have a number, you can bring it, and you can run several numbers on one subscription.
Yes. Instagram DMs can use the same grounded answer and human-handover flow. The public simulator sends nothing; a live connection needs a professional account, the applicable Meta permissions, webhook setup, and the customer-service window. Product-rich content falls back to clear text where Instagram cannot render a native card.
Campaigns never send without your tap. In conversation, answers are grounded in your own pages and products and cite their source, and anything the sources do not cover is handed to a person instead of guessed at.
No. Your ledger belongs to your account. You can also bring your own AI provider keys, in which case the platform runs entirely on them.
30-day free trial to start, and we do not bill per WhatsApp number or per LINE account — one subscription covers the accounts you actually run, staging included. There is no per-message meter and no call to book before you learn the price: the plan is the same for every account, and the checkout shows it before you pay. Managed messaging services in this category quote regional per-message rates on a sales call and bill from a prepaid credit balance, so the month's cost is known only after the month.
Meta restricted AI Providers — companies whose general-purpose AI assistant is the main thing they distribute through WhatsApp — from 15 January 2026, and its terms scope that to where such technology is the primary rather than incidental functionality. That is not what runs on your number here: the AI answers about your own products, orders and bookings from your own data, and hands anything your sources do not cover to a person. Meta's own note on the change says it does not change how Meta charges all other businesses. Rules differ for numbers registered in the European Economic Area and Brazil, so check Meta's current terms for your market.
A number gets banned for sending what Meta did not approve, to people who did not opt in. Here, campaigns can only go out on templates Meta has already approved, marketing consent is recorded in a ledger you can inspect, and the send gate fails closed when that consent is missing or the recipient is in quiet hours. Because every send is one you approved, in your own console, the number carries your history and no one else's. Managed messaging services in this category send on your behalf from a shared operation, so a ban that lands on your number is the result of a decision you never saw.
Anywhere WhatsApp Business is available, which is most of the world. The product also speaks LINE, which matters in Japan and Thailand, and the website chat works everywhere.
TickTock Town (ticktock-town.com) is the live demo town that runs on DigiClient AI: AI characters operate real little shops there, and AI agents are welcome to visit them over MCP. The platform behind the town — and behind this site — is DigiClient AI.
30-day free trial* — try it against your real customers.