Wonderchat
Wonderchat builds AI support and conversion agents trained on your own help centre, documents and resolved tickets. Every reply cites its source, low-confidence questions escalate to a human, and captured leads sync straight into your CRM.
What is Wonderchat?
Wonderchat is a customer-facing AI agent platform published by Wonderchat Pte Ltd in Singapore and online since 2023. One agent covers three commercial uses: a support agent that deflects repetitive tickets, a conversion agent that qualifies visitors and captures leads, and an internal agent that answers staff questions from company documents. The product is built around grounded retrieval. The agent answers only from content the business has approved, and every reply cites the page or file it came from. Knowledge can come from a crawl of your website, private crawling behind a login, uploaded PDFs, documents and spreadsheets, media extraction from webinars and recordings, and resolved support tickets, with scheduled sync keeping everything current. A QA agent flags weak answers for one-click correction, and each correction feeds back into retrieval so the same gap does not reopen. Every answer carries a confidence score, and fallback detection hands the conversation over rather than guessing. Coverage is broad. Wonderchat claims responses in more than forty languages with no per-market setup, and lets you choose the underlying model among OpenAI, Anthropic's Claude, Google Gemini, Llama on Groq, a US-hosted DeepSeek model and Perplexity, switchable at any time. The same agent deploys to a website widget, an AI search bar, Microsoft Teams, Slack, Discord, Messenger, Telegram, WhatsApp, SMS through Twilio and voice calls. A documented REST API, a client SDK and webhooks are available for deeper work. The publisher advertises 500+ teams onboarded, over two million tickets resolved, a resolution rate above 70% and a 4.9/5 average rating; its support pages show a trailing three-month resolution rate rising from roughly 54% to roughly 76% after multi-pass retrieval and the QA loop shipped in late 2025. Named customers include Ricardo.ch, Jortt, Ko-fi, 10xTravel, Filewave, Trend Micro, Korendy and Yale's Office of International and Summer Programs, with Jortt reporting 92% of repetitive tickets deflected. Pricing deliberately rejects per-resolution metering in favour of a monthly credit allowance, and compliance is a stated priority: SOC 2 Type II, GDPR, EU hosting in Frankfurt, a HIPAA BAA, SSO and role-based access on the upper tiers.
What it does
- Answer repetitive customer questions from your own documentation, citing the source in every reply
- Escalate to a human, an email or a helpdesk ticket whenever confidence in the answer drops
- Qualify website visitors, capture their details inline and push the lead into your CRM
- Train an agent on a website crawl, uploaded files, gated pages and resolved support tickets
- Book meetings against live Calendly availability without leaving the conversation
- Deploy the same agent on your site, Microsoft Teams, Slack, WhatsApp, Telegram, Discord, SMS or voice
- Surface what the agent could not answer through gap analysis, FAQ clustering and resolution reporting
When to use Wonderchat / When not to
A quick filter to help you decide if Wonderchat is the right fit.
When to use Wonderchat
- Support teams drowning in repetitive tickets who already have a help centre, product docs or a ticket history to train on
- B2B and e-commerce websites that want to qualify visitors and capture leads inside the conversation rather than behind a form
- European or regulated organisations that need EU hosting, SOC 2 Type II and a HIPAA BAA on the table
- Teams tired of per-resolution billing who want a predictable monthly allowance instead of a meter that ticks on every answer
- Agencies and multi-brand operators who need to move corrections and configurations between agents and keep client workspaces separate
When not to use Wonderchat
- Organisations with no documentation to point the agent at, since it answers only from the content you supply
- Buyers looking for a general-purpose conversational chatbot rather than one grounded in their own data
- Tight budgets: the free plan covers roughly ten resolutions a month and the first paid tier is USD 124 per month on an annual commitment
- Procurement teams that require a negotiated DPA, as the publisher will not modify or sign an additional one
- Projects aimed at audiences under 13, which the acceptable use policy explicitly forbids
How to use Wonderchat
A typical end-to-end flow, from setup to results.
- Create an account at app.wonderchat.io/try, with no credit card required
- Point the agent at your domain so it crawls your pages, and upload the PDFs, documents and spreadsheets it should answer from
- Connect your knowledge systems such as Microsoft SharePoint, Google Drive, Confluence or your helpdesk, and add resolved tickets as training material
- Let the initial training run; the publisher advertises roughly five minutes for a website and its files
- Set the persona: name, avatar, tone, custom instructions, greeting and suggested prompts
- Configure escalation: confidence threshold, fallback detection, live chat, email escalation and helpdesk hand-off
- Publish the agent by pasting the embed snippet on your site, or activate a channel such as Teams, Slack, WhatsApp, Telegram, Discord or Messenger
- Turn on lead capture and connect HubSpot, ActiveCampaign, Salesforce or Microsoft Dynamics 365 so qualified leads land in your CRM
- Review flagged answers, correct them once, and let the correction feed back into retrieval
- Track resolution rate, satisfaction, FAQ clusters and gap analysis in the dashboard, or receive automated reports by email
Pros & Cons
Pros
- Every answer cites its source, which makes verification possible instead of a matter of trust
- Fallback detection and confidence scoring push the agent to hand off rather than invent
- A serious compliance file for a company this size: EU hosting in Frankfurt, SOC 2 Type II, a named sub-processor list and a HIPAA BAA
- Flat monthly pricing rather than per-resolution metering, with unit cost falling from roughly USD 0.12 to USD 0.08 per resolution
- Wide channel and integration coverage, including WhatsApp, Microsoft Teams, SharePoint, Zendesk, HubSpot and Shopify
- The underlying AI model is a customer choice and can be switched at any time
- Detailed public documentation with an API reference and nineteen integration guides, plus named and quantified case studies
Cons
- High entry point: USD 124 per month on an annual commitment, while the free plan covers only about ten resolutions a month
- Month-to-month pricing is injected in JavaScript and never appears in the static page, so the headline figures assume an annual commitment
- The free trial is advertised as seven days on the pricing page and fourteen days on the competitive landing pages
- Deletion timing is stated as seven days in the privacy policy and fourteen days on the GDPR page and compliance portal
- The standardised DPA cannot be negotiated, and the terms of service exclude refunds altogether
- No option is documented for excluding customer data from model training
- No about page, no team page and no legal notice: the publisher's identity appears only inside the terms and the privacy policy
Pricing & Plans
Wonderchat offers a permanent free plan at USD 0 per month, limited to roughly ten resolutions, twenty credits and fifty stored web pages. The cheapest paid tier is Basic at USD 124 per month billed annually. A seven-day free trial is available without a credit card, and annual billing is advertised as 17% cheaper than monthly payment. The terms of service exclude refunds.
- about 10 resolutions per month
- 20 credits
- 50 stored web pages
- 1 AI agent
- human handover
- basic integrations
- embeddable chat widget and file upload
- about 1
- 000 resolutions per month
- 5
- 000 credits
- 1
- 000 web pages per agent
- 2 AI agents
- 3 team seats
- about 5
- 000 resolutions per month
- 25
- 000 credits
- 2
- 500 web pages per agent
- 4 AI agents
- advanced analytics
- custom resolutions and storage
- bulk message discounts
- priority support
- SOC 2 and HIPAA
- custom integrations
- SSO and an SMS agent
- Add-on
- USD 29.99 one-off for an extra 1
- 000 message credits
- purchasable from the dashboard
Data, GDPR & hosting
A consolidated view of how Wonderchat handles your data.
GDPR overview
GDPR compliance is claimed explicitly and backed by concrete detail. Servers and databases sit in Frankfurt, Germany, and the publisher states that no personal data leaves the EU. Data is encrypted at rest and in transit over TLS, and passwords are not stored. A standardised DPA is available on request from hello@wonderchat.io, though it cannot be modified or replaced. A named list of seventeen sub-processors is published on the compliance portal. Breaches are reported to the supervisory authority within 72 hours, with the Irish Data Protection Commissioner named as the point of contact. Access, rectification, erasure, restriction, portability and objection requests are handled within one month after identity verification, the first copy free of charge. An internal GDPR owner is designated. No Article 27 EU representative is named anywhere on the site.
Who owns the data?
Wonderchat draws a clear line. For anything a chatbot collects from visitors, the customer who built that chatbot is the data controller and Wonderchat is only the processor; the publisher states plainly that it provides the service and does not own the collected data. For account data such as registration, billing and usage, Wonderchat acts as controller. Respondents who want their answers removed must approach the business that built the bot, not Wonderchat. Customers keep operational control throughout: they set their own retention period, and they can delete or export chat data and entire chatbots from the dashboard at any time.
Reuse rights
Wonderchat collects electronic identifiers such as IP address, device and browser data and location, along with site usage, registration details, billing information held by its payment provider Paddle, chatbot data covering the knowledge base, files, chat logs and analytics, and anything pulled from connected third-party integrations. It states that this data serves only to deliver and improve the service, send essential product updates, fix bugs, answer legal requests and prevent fraud. Personal data is never sold and never used for marketing or advertising, and under Article 28 it is processed only on the customer's instructions and never passed to a third party without consent. Data drawn from third-party APIs such as the Google Workspace API is explicitly excluded from developing, improving or training generalised AI and ML models. Access is limited to the technical team responsible for server stability and is logged in an activity trail. Customers may reuse and export their own data freely, in industry-standard formats.
Data retention & training
Hosting summary
Wonderchat states that its servers and databases are located in the European Union, in Frankfurt, Germany, and that no personal data is transferred outside the EU. Application hosting is provided by Render, whose Frankfurt servers are named on the compliance portal, and the database layer by Supabase. Cloudflare handles delivery and network protection, which is why the domain resolves to a US anycast address even though the data itself is described as German. Data is encrypted at rest and in transit over TLS, and passwords are not stored. Backups cover the whole customer dataset, are held for fourteen days and exist for disaster recovery only. Seventeen sub-processors are named publicly, several of them outside the EU: Paddle in the United States, BentoNow in Australia, Cohere in Canada and Pusher in the United Kingdom, alongside OpenAI, Anthropic, Google Workspace, DocuSign, Retool, Fireflies AI, PostHog EU, Postmark and Apilayer in Germany. That list sits somewhat awkwardly against the no-transfer claim and is worth clarifying with the vendor.
Things to keep in mind
Risks and trade-offs to weigh before adopting Wonderchat.
- An agent that answers from your documentation is only as accurate as that documentation: stale or contradictory pages will be quoted back to customers with full confidence
- Deflection targets can quietly become deflection walls, so measure how easily a frustrated visitor still reaches a human, not only how many tickets disappeared
- The free trial is advertised as seven days on the pricing page and fourteen days on the competitive pages; confirm the real duration before committing
- The privacy policy says deleted data leaves backups within seven days, while the GDPR page and compliance portal say fourteen; ask which one applies to you
- The publisher states that no personal data leaves the EU, yet the sub-processor list names Paddle, BentoNow, Cohere, Pusher, OpenAI and Anthropic, and the feature pages describe a US-hosted DeepSeek model
- No opt-out from model training is documented anywhere, and the statements that do exist are narrower than a blanket commitment
- Teams that hand the whole front line to an agent risk losing the informal signal support conversations carry: a gap analysis dashboard replaces, but does not equal, someone reading the tickets
Setup & Integrations
Technical difficulty
Low for a standard deployment. Wonderchat is a no-code product: you point it at your domain, upload files, and paste an embed snippet that works on Framer, Webflow, WordPress, Shopify and custom builds. The publisher advertises roughly five minutes to a working agent. Connectors handle helpdesk, CRM and messaging channels. Technical skills only become necessary for the advanced layer, meaning the REST API, client SDK and webhooks, private crawling behind authentication, CSS overrides and custom databases, and for SSO and role-based access control, which are reserved for the upper tiers.
Deployment
Integrations
Behind Wonderchat
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Wonderchat.
Frequently asked questions
What does Wonderchat train on?
Does every answer show where it came from?
What happens when the agent does not know?
Is there a free plan?
How much does the cheapest paid plan cost?
Where is customer data hosted?
Is a data processing agreement available?
Which certifications does Wonderchat hold?
Is there an API?
Is there a mobile app?
Should you pick Wonderchat?
Wonderchat is a credible option for teams that already own a body of documentation and care about being able to check what an AI told their customers. Grounded retrieval with a citation on every reply, confidence scoring and fallback detection are the real differentiators here, and they address the failure mode that has made many buyers wary of support chatbots. The correction loop, which pushes a fix back into retrieval rather than leaving it as a one-off patch, reads as an engineering choice rather than a marketing line. The compliance file is unusually solid for a company of this size: servers in Frankfurt, SOC 2 Type II, a named list of seventeen sub-processors, a 72-hour breach process and a HIPAA business associate agreement on the enterprise tier. European buyers will find most of what procurement asks for, with two caveats: the DPA is standardised and non-negotiable, and Singapore law governs the contract. Pricing is honest about its logic and expensive at the door. Refusing per-resolution metering is a defensible position and the unit cost genuinely falls as volume grows, but USD 124 per month on an annual commitment is a real commitment, and the free plan is a demonstration rather than a working tier. The weakest part is corporate transparency. There is no about page, no team page and no legal notice; the publisher's identity has to be pieced together from the terms and the privacy policy, which do not even write the company name the same way. Two documented figures also contradict each other, trial length and deletion delay, and neither is resolved anywhere on the site. None of this is disqualifying, but it is worth clarifying with the vendor before signing.
- Choosing a selection results in a full page refresh.
- Opens in a new window.