GaliChat

A one-sentence summary

GaliChat is a code-free AI customer service and sales bot designed for corporate websites; it can learn from the company’s web pages, documents, and manual materials to answer visitors’ questions, gather leads, and assist with scheduling appointments or making purchases.

Tool Introduction

GaliChat brings together knowledge base training, chat components, lead management, and conversation analysis in a single console. Users do not need to develop their own conversation systems; after completing the training process, they can simply copy the embedding code to integrate the intelligent agent into their website.

This product is more suitable for small and medium-sized enterprises that already have websites, as well as those that have common questions and relevant product documentation. It belongs to the category of hosted business services; it is not an open-source chatbot whose source code can be downloaded for self-deployment.

Main functions

Train the knowledge base using corporate data

When creating an agent, it is possible to capture website links, upload multiple files, or enter text directly. The system uses the selected materials as a basis for generating responses; the training time generally increases as the number of links and files rises.

• Website scraping: Enter the domain name to access the page, after which the user selects the links that are to be included in the training process.
• File upload: Documents in PDF, Word, TXT formats are supported; Excel is also listed as an available file type on the official website.
• Manual text: It can be used to add product descriptions, prices, policies, and FAQs.
• Additional training: When the response is incomplete, links, files, or specific text can be added.

Website AI customer service Q&A

Once the training is complete, the agent can continue to answer visitors’ questions in the web chat window. The quality of these responses depends on the availability of relevant data, the prompts used, the model selected, and the way in which the visitors formulate their questions; therefore, the accuracy figures listed on the marketing pages should not be considered as a guarantee of the outcome.

Clue collection and export

Administrators can specify the names, email addresses, or phone numbers that need to be collected, and they can view and export the relevant records on the Leads page by date. Before collecting this information, it is necessary to explain its purpose to the visitors and obtain their consent in accordance with local regulations.

Conversation history and continuous improvement

On the Messages page, it is possible to view the messages sent by both visitors and agents, thereby identifying knowledge gaps in response to actual questions. The official introductory documentation states that messages are automatically synchronized every 10 minutes; it is also possible to manually refresh the current session.

Brand and appearance customization

Users can set the name, color, avatar or logo of the agent, as well as choose between dark or light mode. Removing the GaliChat brand and using a custom brand involves either benefits included in the package or additional fees.

Prompt and model settings

The settings page enables the modification of the agent’s prompt and the GPT version, allowing adjustments to the scope of responses, the role assigned to the agent, and its way of expressing itself. The version names and available options may change, so it is necessary to refer to the current console for accurate information.

Multilingual customer service

The paid plan includes multilingual capabilities; the official website states that it supports more than 50 languages. The free trial is limited to English only, so teams that need customer service in Chinese should first try out the paid features using actual business-related content.

Cal.com booking integration

The pricing plans for Pro, Business, and Enterprise all include integration with Cal.com, which allows visitors to proceed to the booking process within a conversation. The actual available booking times, notifications, and data processing are still determined by the settings of the associated service.

Sharing and website embedding

Agents can be shared via a separate link, or a piece of code can be copied to embed them in a website’s HTML. The official website focuses on website components and does not provide verifiable native clients for Windows, macOS, iOS, or Android.

The workflow from data to going live

1. Create an account and go to the My AI Agents page.
2. Create a new agent and choose website scraping, file uploading, or manual text entry.
3. Check the captured links and retain only those pages that are related to the visitor’s question.
4. Start the training and wait for the data processing to be completed.
5. On the preview page, test using common questions, edge cases, and error scenarios.
6. Fill in the missing information, and adjust the prompts, models, and style of the responses.
7. Configure the fields of information to be collected and the privacy notice.
8. Copy the embedded code to your website and observe the actual conversations.
9. Regularly review messages, update the knowledge base, and remove outdated materials.

Tutorial on creating the first AI customer service system

1. Register and log in using your email password or Google account.
2. Click to create an AI Agent, and enter a recognizable name for it.
3. Enter the company’s website domain to scrape the page, or upload the relevant documentation.
4. Check the key pages such as products, prices, returns and exchanges, shipping, and contact information.
5. After the training, tests are conducted on simple questions, multiple follow-up questions, and questions related to information outside the training material.
6. Provide additional clarifying information for incorrect answers; do not rely solely on more complex prompts.
7. After verifying the accuracy, obtain the embedding code and publish it to the testing environment.

Tutorial on configuring lead collection

1. First, identify the contact fields that are truly necessary for the business, adhering to the principle of collecting only what is essential.
2. In the agent settings, enable the clue feature and select name, email, or phone number.
3. Specify clearly the purpose of data collection, the methods for subsequent communication, and the privacy policies.
4. Simulate both the submission and rejection scenarios by visitors to ensure that normal Q&A operations are not disrupted.
5. Check the records on the Leads page, and test date filtering and export.
6. Restrict team access rights and delete unnecessary data according to the internal retention schedule.

Tutorials for improving the quality of responses

1. Organize frequent questions and unresolved issues on the Messages page.
2. Verify that the response indeed comes from the most up-to-date company information.
3. Delete conflicting, duplicate, and invalid links or files.
4. Write direct, unambiguous, and conditional text for important rules.
5. Adjust the prompt to specify how to respond when something is unknown and when to transfer the request to a human agent.
6. After retraining, a fixed set of test questions is used to compare the results before and after.
7. Manual review is retained in cases involving medical, legal, financial matters, or contractual commitments.

Which users are it suitable for

• Websites for small and medium-sized enterprises: Automatically answers questions regarding business hours, services, and policies.
• E-commerce team: Explains product details, shipping and return policies, and guides users to the product pages.
• SaaS team: Uses documents and frequently asked questions to handle pre-sales inquiries and basic support.
• Course and consultation services: Answering questions about the range of services offered and assisting visitors in making appointments.
• Marketing team: Collects the contact information of potential customers and analyzes the most frequent requests.
• Content creators: Create Q&A sections around courses, podcasts, books, or events.
• Small customer service team: Utilize round-the-clock self-service Q&A to reduce repeated inquiries.

Typical use cases

• The product’s official website answers questions regarding features, prices, and common issues before purchasing.
• Add shipping, return, and warranty policies to the knowledge base to handle inquiries related to orders.
• Collect visitors’ names and email addresses and pass them on to the sales team for further follow-up.
• Use Cal.com to direct highly interested visitors to available booking times.
• Answer questions regarding the content of the course website, its target audience, and the registration process.
• Apply the brand’s tone and avatar to the website’s chat components.
• By analyzing the conversation history, it was found that there are no issues in the page content that require clarification.

Product advantages

• There is no need to develop a complete chat system; it is suitable for companies that do not have an AI engineering team.
• The three training methods – websites, files, and text – cover common sources of knowledge input.
• Centralize Q&A, clues, conversation history, and appearance management in the console.
• The paid plan supports multiple languages and integration with Cal.com for scheduling.
• You can first use a free trial that requires no credit card to verify the basic process.
• The chat component can be embedded in existing websites, or an independent sharing link can be used.
• The package limits are listed explicitly for agents, links, messages, leads, and files.

Usage restrictions and precautions

• The free trial allows only 40 messages and is available in English only; it is not equivalent to the full free version.
• After 7 days with no activity in the free trial, the chat function will be disabled.
• AI responses may be incorrect, incomplete, or misinterpret the information; the official terms do not guarantee accuracy.
• Paid plans also have limits on the number of messages, links, leads, and files.
• The efficiency and accuracy figures shown on the official website’s marketing pages lack publicly available testing methods, so they cannot be used as a guarantee for purchases.
• Embedding a website requires modifying the page code or adding scripts through a website construction platform.
• Training data is sent to external AI model services for processing, while sensitive files must first undergo a compliance assessment.
• Message history, leads, and exported files may all contain personal information, and therefore require appropriate permissions as well as retention policies.
• Subscriptions are prepaid on a periodic basis, and within the 14-day refund period following the initial purchase, the unused portion is generally not refunded.
• The authorities do not provide any verifiable self-hosted versions; organizations with strict requirements regarding data storage should consult them first.

Prices and packages

The price verification date is August 22, 2026. The official website offers the option to pay monthly or annually; it states that annual payment saves 20%, but the currency used for settlement, any taxes, and the final amount shall be as indicated on the purchase page.

Common capabilities and differences of the packages

Price for additional features

Refunds and cancellations

The terms stipulate that the subscription fee must be paid in advance for each billing cycle. A refund can be requested within 14 days of purchasing the first subscription, as long as no violations of the terms have occurred; cancellation after that usually takes effect at the end of the current cycle, and no refund is issued for unused periods.

Support for platforms and input/output.

APIs, SDKs, and open-source status

The privacy policy and terms of service list the API as part of the services, indicating that the platform offers interfaces for system integration. However, the official website does not provide sufficient developer documentation, information on authentication methods, pricing details for API calls, or a list of official SDKs.

As of the time of verification, no official open-source repository that can be identified as belonging to the GaliChat product was found. The product should be classified as a closed-source commercial service; it cannot be labeled as open source just because its page assets are stored on GitHub or because it uses third-party open-source components.

Privacy and data processing

According to the privacy policy, account and access data may include name, email address, contact information, IP address, browser type, pages visited, device identifiers, and Cookie data. Billing and payment information may be retained for 10 years as required by applicable regulations.

The platform relies on AWS, Supabase, and Vercel for hosting or storage, Paddle handles payments, and the OpenAI API is used to process training data. According to the policies, only text, PDFs, documents, and information provided by the owners of the agents are sent to the model services; personal data or detailed information about the owners is not sent to the model APIs.

• Remove identification documents, health records, contract secrets, and unnecessary personal information before uploading.
• Specify the purpose of collection, retention period, and contact information in the form for providing clues.
• Lead and conversation data are available only to team members who need to follow up.
• Pay attention to cross-border data processing, third-party processors, and the data regulations in the respective regions.
• Users may request access to, correction of, deletion of, restriction on, objection to, or transfer of their personal data.
• Closing an account may result in the loss of access to content and services; it is therefore necessary to export the relevant business records first.

Basic information

Recommendation score

Recommendation score: 4.2 / 5. GaliChat is suitable for small and medium-sized teams that wish to quickly turn their business information into a website customer service system, while also collecting leads and handling reservations; its interface is user-friendly and the pricing plans are transparent.

The main shortcomings are the limited amount of free trial credits, incomplete API documentation, the absence of an open-source self-hosting option, and the fact that training data is processed by third-party model services. Before making a purchase, it is essential to test the capabilities related to Chinese-language responses, knowledge base updates, and the handling of sensitive data.

Frequently Asked Questions

Is GaliChat free?

A free demo is available without the need for a credit card; it includes 1 agent, 5 training links, 40 messages, up to 5 clues, and 20 files. It is only available in English, and the chat will be discontinued if no activity takes place for 7 consecutive days.

Does GaliChat support Chinese?

The paid plan includes multilingual capabilities; the official website claims support for over 50 languages, but it does not provide a detailed list of these languages. Chinese terms, long documents, and multi-turn conversations should be tested using actual text samples before purchase.

How to add an agent to a website?

After training and testing are complete, copy the embedded code from the console and paste it into the website’s HTML. When using WordPress, Shopify, or another website building system, it needs to be placed in the area where scripts are allowed to be inserted.

What materials can be used for training?

It is possible to capture website links, upload files such as PDF, Word, and TXT, or simply type in text. The actual file size, number of characters, and format compatibility are determined by the console.

Can customer leads be collected?

Fields such as name, email address, and phone number can be configured, and they can be viewed, filtered, and exported on the Leads page. Companies are responsible for providing their own privacy notices and handling the collected information in a lawful manner.

Can I view the chat history?

The messages exchanged between the two parties can be viewed on the Messages page, which helps to identify any knowledge gaps. The official tutorial states that automatic synchronization occurs every 10 minutes; manually synchronizing the data can also be used to view more recent sessions.

Does GaliChat provide an API?

Legal documents state that API is part of the service, but the public website does not provide complete development documentation. Teams that need to integrate this API programmatically should first consult the authorities regarding permissions, pricing, authentication requirements, and any restrictions.

Is GaliChat open source?

No. Currently, no verifiable official source code repository or open-source license can be found; it should therefore be considered a closed-source hosting service.

Can it be deployed privately?

The official website does not provide information on self-hosted or on-premises deployment options. Companies that have requirements regarding data storage, private networks, or offline operation need to contact sales for clarification.

Can a subscription be refunded?

A refund can be requested within 14 days after the initial subscription, in accordance with the terms and conditions, provided that no violations of those terms have occurred. Cancellation after this period usually only prevents renewal for the next cycle; it does not result in a refund for the unused portion of the current cycle.

Are AI responses always accurate?

No guarantee can be provided. Insufficient knowledge bases, conflicting information, and misunderstandings of the models can all lead to errors; the official terms also state that no guarantee is given regarding the accuracy or suitability of the responses.

Will the training materials be handed over to third parties?

Yes. The privacy policy states that training texts and files are sent to the OpenAI API for processing; therefore, a compliance assessment must be carried out before uploading sensitive or regulated data.

Summary

GaliChat combines website content training, AI customer service, lead generation, conversation analysis, brand customization, and booking functions into a code-free workflow, making it suitable for corporate websites that seek to go live quickly.

It is recommended to first use the Free Demo to test content extraction and response capabilities, and then choose a plan based on the volume of messages, links, clues, and files. Before going live, it is also necessary to verify the performance in Chinese, the privacy policy, the mechanism for human intervention, and the ways in which incorrect responses are handled.