Chatty AI

What is Chatty AI?

Chatty AI is a Chinese-based AIGC web service operated by Chatty AI LLC.; it offers services related to writing, learning, programming, social chatting, and guidance on various tasks, through virtual characters and AI-driven conversations at different levels.

The main server and the applications are still accessible; users need to register or log in via email. Unlogged sessions show 0 available uses and prompt the user to purchase additional uses or a monthly subscription, so it is not a free chat tool that can be used indefinitely without an account.

Product positioning and current status

Main functions

Multi-role AI dialogue

• The platform advertises thousands of virtual characters, including IELTS teachers, project managers, diplomats, historical figures, and fictional characters.
• After selecting a role, the user can engage in natural-language conversations, which can be used for knowledge quizzes, scenario-based exercises, the exchange of ideas, and casual chatting.
• Character settings are merely part of the framework for generative dialogue; they do not represent real people, and they cannot guarantee accurate responses regarding a character’s background, the settings of their works, or historical facts.
• When dealing with real people, it is necessary to avoid creating statements that are fraudulent, defamatory, contain private information, or could mislead others.

Writing assistant

• It can handle tasks such as summarization, expansion, weekly reports, naming, drafting replies, creating content for social media platforms, correcting spelling errors, and polishing articles.
• The user enters the topic, target audience, tone, length, and the facts that must be included; the system then returns a draft text that can still be modified.
• Editing and rewriting papers is intended to improve their structure and expression; it should not be used to fabricate research, circumvent academic standards, or conceal improper plagiarism.
• For social and news commentary content, it is necessary to verify facts, dates, and copyrights, to prevent model predictions from being regarded as established facts.

Learning and work assistant

• Learning roles can assist in reading and understanding papers, creating study plans, discussing questions after class, and organizing one’s approach to homework.
• In a work scenario, users can assume the roles of project managers, consultants, or other positions to help break down tasks, prepare for meetings, and create checklists.
• Models may provide seemingly complete but incorrect explanations; students should check the textbooks and course requirements, while professionals in the workplace should verify company policies and actual data.
• AI-generated work should not be submitted directly, nor should fictional cases, data, or citations be included in formal submissions.

Code Assistant

• The product page lists uses such as SQL, front-end components, RESTful APIs, operation and maintenance scripts, code analysis, comments, and test case generation.
• Users can provide the language, framework, runtime environment, input/output data, and error messages, so that the system can generate code and make adjustments based on those feedbacks.
• The page claims a high accuracy rate, but since no test set, evaluation method, or version details are provided, this figure cannot be considered as an indicator of quality.
• The generated code must undergo testing, security reviews, and dependency checks, with a particular focus on preventing injection attacks, excessive permissions, key leaks, and destructive commands.

Process and inquiries

• The characters can provide general steps, lists of required materials, and simulated scenarios related to visas, studying abroad, licenses, tendering processes, corporate affairs, and business procedures.
• Users should specify the country, region, identity, time of processing, and the purpose of the request, so that the system can provide more appropriate preliminary suggestions tailored to the specific situation.
• Laws, immigration rules, licensing requirements, and bidding regulations can change; AI does not have the authority to act as an agent, and formal applications must comply with the requirements set by the relevant authorities and licensed professionals.
• The platform has not demonstrated that its personnel possess qualifications as lawyers, immigration consultants, or other regulated professionals; therefore, it is not suitable for making high-risk decisions.

Document-based input and real-time preview

• The editor supports multi-line Markdown input, allowing for the organization of headings, paragraphs, tables, formulas, code, and links, with real-time preview of the layout.
• Custom send keys help reduce accidental sending due to line breaks, and are suitable for writing longer prompts, technical questions, and structured task instructions.
• This “document-based dialogue” refers to rich-text input; it is not the same as uploading PDFs, Word documents, or spreadsheets for file processing.
• The maximum number of characters per conversation within the package limits the length of text that can be entered; very long texts must be divided into sections, with clear section numbers included.

Model and Intelligence Level

The product page lists open-source NLP models such as GLM-130B, Vicuna, and Chinese-LLaMA-Alpaca, and categorizes the packages according to AI intelligence levels I, II, and III. The page does not disclose the specific models associated with each level, nor the parameters involved, the context length, or the routing rules.

• The fact that a page lists a certain open-source model does not mean that that model will be used in every conversation, nor does it imply that users can select or download the model.
• GLM-130B, Vicuna, and Chinese-LLaMA-Alpaca all belong to the earlier generation of models; before making a payment, it is necessary to check with customer service to determine which version of the model is currently in use and what the update policies are.
• The “Intelligent Level” tier is a custom classification defined by the product itself; it is not an industry-standard capability, and therefore cannot be directly compared to the tier levels of other services.

Session storage and multi-device synchronization

• Chatty AI offers cloud-based session storage, along with local storage in the browser, to enable continued work on phones, tablets, and computers.
• Silver corresponds to 10GB of cloud backup and the migration of 1000 conversations; Gold corresponds to 20GB and 4000 conversations; Prime offers unlimited capacity for both aspects.
• The secondary card indicates 100 entries of historical data, but it does not specify whether this refers to data stored in the cloud, on the current device, or as a total for each account.
• The page does not disclose information regarding synchronization conflicts, export formats, methods of deletion, the retention period for backups, nor the boundaries of fair use in the case of \"unlimited\" usage; important conversations should be saved separately.

Input and output

Usage tutorial

1. Select registration in the web application to create an account using an email address; if you already have an account, log in directly.
2. Go to the purchase page, first select whether to opt for a single card or a subscription based on the frequency of use; before making the payment, take a screenshot to record the actual amount, duration, and benefits.
3. Create a new dialogue and select the appropriate role; describe the background of the task, its objectives, the target audience, the desired output format, as well as any facts that must not be invented.
4. For long prompts, use Markdown to divide them into headings, requirements, materials, and examples; be mindful of the maximum character limit for a single prompt under the current plan.
5. First, have the AI generate an outline or a list of steps, and then add details to each item one by one; for coding tasks, it is also necessary to provide the execution environment as well as any errors that can occur and be reproduced.
6. Manually verify numbers, facts, policies, quotes, and code; do not copy them directly into official documents or the production environment.
7. Before using across devices, check whether the session has been synchronized; export or save important results separately, and do not rely on cloud chat as the only backup.
8. When it is no longer in use, clear the data from the local browser and contact customer service to confirm how the data will be deleted in the cloud, the subscription will be terminated, and what will happen to any remaining usage credits.

Price of secondary cards

The secondary cards are charged based on the number of conversations that can be sent; all three tiers utilize level I intelligence, a limit of 500 characters per message, 100 entries for storing historical messages, and multi-device support. The page does not specify the validity period of this usage quota, nor whether failed requests count towards the quota or if refunds are possible.

Subscription price

Subscriptions are categorized by intelligence level, number of characters per submission, cloud backup, and historical migration. The page title indicates discounts for monthly, quarterly, and annual plans, but some plans show two different amounts, and these figures do not always match the discounts stated in the title.

• The two amounts shown on the quarterly and annual cards may represent the original price and the discounted price; however, the text on the page does not make it possible to determine clearly which amount is which, so the actual amount paid on the settlement page should be taken as the reference.
• The public page does not indicate how many conversations are included in the subscription, whether renewal is automatic, what happens to cloud backups once they expire, or the different levels of response priority.
• “There is no set of rules for fair and open use regarding ‘infinity’; it cannot be interpreted as meaning that there are no restrictions on frequency, duration, or permanent storage.”
• No clear rules regarding refunds and cancellations were found; before purchasing a seasonal or annual pass, it is necessary to obtain written confirmation from customer service regarding the unused period, the remaining number of uses, and the procedures for dealing with service issues.

Supported platforms and languages

API, GitHub, and open-source status

• No public API, SDK, development documentation, official product repository, or open-source license for the Chatty AI product was found; therefore, its web service should be considered a closed-source hosted product.
• The Chinese-LLaMA-Alpaca repository mentioned in the database records is one of the third-party model options listed on the product page; it is not part of the Chatty AI client or server code.
• The code for this third-party repository is licensed under Apache 2.0, but the model weights and related resources are subject to separate licensing terms; they are intended solely for academic research and cannot be used for commercial purposes.
• Open-source model components do not prove that Chatty AI as a whole is open source, nor do they automatically grant users a commercial license to use the services, model outputs, or derived content.
• Teams that need local deployment, batch calls, or integration into business systems should first check with the service provider to determine whether unpublished enterprise APIs and written authorization are available.

Privacy and security

The product page states that users have control over their data, with cloud and local storage used in combination, but no public privacy policy, security guidelines, data processing protocols, or retention periods were found.

• Registration requires an email address, and the contact form may also collect the name, phone number, email address, and a description of the issue; however, it is not specified where these fields are stored or how they can be deleted.
• The session is stored in cloud backup and used for synchronization across multiple devices; just because local storage also exists, it does not mean that the content remains only on those devices.
• The page does not provide information on transmission and static encryption, employee access control, processing by third-party models, server location, backup frequency, and notifications for security incidents.
• Do not enter identification documents, payment information, account passwords, API keys, customer data, unpublished code, or materials subject to confidentiality obligations.
• Enterprises or organizations in regulated environments should obtain written instructions regarding data processing, deletion, subcontracting, and cross-border transmission prior to making purchases; otherwise, such approaches are only suitable for tasks with low sensitivity.

Copyright, commercial use, and refund guidelines

• No public terms of service were found that clarify the ownership or scope of permissions regarding user inputs, AI outputs, role assignments, and feedback.
• Virtual characters may involve names, visuals, trademarks, and narrative elements; before using them publicly or for commercial purposes, it is necessary to assess the risks related to personality rights, trademark rights, and copyright.
• The written text and the code output may be similar to existing content or may contain incompatible code; it is necessary to check the copyright, licenses, and third-party dependencies before publishing.
• The third-party Chinese-LLaMA-Alpaca model project explicitly restricts the commercial use of its model resources, but Chatty AI does not disclose its actual model composition or the framework for commercial licensing.
• The rules regarding refunds, automatic renewals, package changes, expiration of usage limits, and account termination have not been made public yet; it is necessary to save the order before making a payment and obtain confirmation through after-sales support.

Suitable for users and typical use cases

• Individual writers and operators: create outlines, short articles, weekly reports, and draft social media posts.
• Students and self-learners: Explaining key concepts, breaking down essays, and creating study plans, while also checking course materials on their own.
• Developer: Generates SQL statements, front-end code snippets, API examples, scripts, and test cases, which are then tested in an isolated environment.
• Language learners: Practice expressing ideas in both Chinese and foreign languages with teachers or role-play partners, without using the feedback from these partners as a standard for grading.
• Users handling everyday tasks: They organize standard procedures and lists of required documents, but this does not replace professional services in the fields of law, immigration, or licensing.
• Entertainment users: Engage in creative conversations with historical figures, celebrities, anime or game characters, while avoiding misleading or infringing content.

Advantages and limitations

Summary

Chatty AI is suitable for individual users who wish to carry out writing, learning, coding, and handling everyday questions quickly using Chinese-language interfaces; its Markdown editing function, ability to sync data across multiple devices, and option to select different languages make it quite useful.

The information regarding its page model is outdated; the relationship between quarterly and annual amounts, the number of subscriptions, refunds for renewals, data policies, and commercial licensing are not clear enough. It is recommended to purchase a small number of cards first to test its performance, and to assess the actual model, settlement amounts, and data requirements before considering long-term packages.