In the AI era, how to use AI tools to do user research

This article will delve into how AI empowers the entire process of user research, from research planning, execution, data analysis to report generation, comprehensively improve efficiency and depth, and conduct an in-depth analysis of several core AI user research tools, while looking forward to the future development trend of AI user research, providing valuable insights and operational guidance for product managers and user research practitioners.

1. Introduction: New opportunities for user research under the wave of AI

1.1 The importance of user research: A deep understanding of users is the cornerstone of product success

In the product-driven era, user research is the bridge connecting products and users, and is the core link to explore real needs, verify product value, and drive innovation iteration. Only by truly understanding users can we create popular products.

1.2 Challenges of traditional user research: efficiency, cost, depth, etc. are becoming increasingly prominent

Traditional user research methods, such as focus groups, in-depth interviews, and large-scale questionnaires, have value but face challenges in terms of efficiency, cost, sample coverage, and depth of insight. Data collation and analysis are time-consuming and labor-intensive, excellent research talents are scarce and costly, and the representativeness of small-sample studies is often questioned.

1.3 The changes brought by AI to user research: intelligence, automation, and scale

The rapid development of artificial intelligence (AI) has revolutionized the field of user research. With its powerful data processing capabilities, pattern recognition capabilities, and natural language understanding capabilities, AI technology is pushing user research to new heights of intelligence, automation, and scale.

1.4 Purpose of the article: Explore the application value, core tools, and future trends of AI in user research

This paper aims to systematically sort out the application methods and values of AI in the whole process of user research, deeply analyze the characteristics and application scenarios of several core AI user research tools, and look forward to the future development trend of AI user research, so as to provide valuable insights and operational guidance for product managers and user research practitioners.

2. Pain points and challenges of traditional user research

In the process of its implementation, the traditional user research model is often accompanied by some unavoidable pain points:

To achieve these three challenges, product managers will only continue to appreciate
Good product managers are very scarce, and product managers who understand users, business, and data are still in demand when they go out of the Internet. On the contrary, if you only do simple communication, inefficient execution, and shallow thinking, I am afraid that you will not be able to go through the torrent of the next 3-5 years.

View details >

2.1 Inefficiency

• Manually designing questionnaires and interview outlines is time-consuming and labor-intensive: The wording, logical order, and avoidance of guidance of the questions all need to be carefully considered.
• Data collection, collation, and transcription are cumbersome: A large number of interview recordings, transcriptions, questionnaire data entry, and cleaning took up a significant amount of researchers’ time.
• Massive qualitative data analysis is difficult and time-consuming: Extracting effective insights from tens of thousands of words of interview records is like looking for a needle in a haystack.

2.2 High cost

• Sample access is difficult, especially for specific groups: Recruiting qualified respondents often comes with high costs and time costs.
• Personnel input (researchers, analysts) costs: Professional research teams and analysts are the guarantee of high-quality research, but it also means high labor costs.
• Cross-regional and cross-cultural research and implementation are complex: Organizing and implementing multi-regional and multilingual research projects has exponentially increased the cost and management difficulty.

2.3 Sample and representative limitations

• Traditional methods are difficult to reach large-scale users: Due to cost and execution capabilities, traditional research is often difficult to achieve large-scale user coverage.
• The conclusions of the small sample survey may be biased: Based on the conclusions drawn from small samples, their universality and representativeness may be insufficient.
• Dynamic user behavior is difficult to capture in real time: User behavior and attitudes are dynamically changing, and traditional research methods are difficult to track in real time and continuously.

2.4 Insufficient depth and objectivity of insight

• Manual analysis is susceptible to subjective bias: The researcher’s personal experience and cognition may unconsciously affect the objectivity of the analysis results.
• Difficult to spot hidden patterns in complex data: Human analysts may struggle to spot deep correlations and patterns when working with highly complex, multi-dimensional data.
• Poor timeliness and missed market windows: Long research cycles can lead to lagging insight results and missing opportunities to respond quickly to the market.

3. How AI empowers the whole process of user research

The integration of AI technology is reshaping every aspect of user research, from planning to reporting, comprehensively improving efficiency and depth.

3.1 Research and planning stage

• AI assists in setting research goals and scopeBy analyzing historical data, industry reports, and market trends, AI can intelligently recommend research directions and assist in defining clear research objectives and scope.
• AI optimizes questionnaires and interview outlines: AI tools can automatically generate preliminary questionnaire questions or interview outlines based on research objectives, and even optimize question wording and logical order based on semantic understanding to reduce human bias. Some AI-powered survey tools, such as SurveyMonkey Genius, can quickly generate high-quality first drafts of surveys.
• AI predicts and filters target users: Combined with multi-dimensional information such as user behavior data and CRM data, AI can assist in user stratification, more accurately locate and screen target respondents, and improve recruitment efficiency and sample relevance.

3.2 Research and implementation stage

• Automated questionnaire distribution and intelligent interactionAI-powered questionnaire tools not only automate distribution, but also dynamically adjust follow-up questions based on users’ real-time responses, and even conduct interactive surveys in the form of chatbots, improving user engagement and data quality.
• Interview assistance and real-time transcription: AI speech recognition technology can transcribe interview recordings into text in real time or quickly with high accuracy, and can distinguish between different speakers, greatly freeing the hands of researchers. Tools like Gong already have such capabilities.
• Multimodal data acquisitionAI supports the integration and analysis of various types of data from different channels, such as text, speech, images, and even user behavior data, to form a more comprehensive user understanding.

3.3 Data analysis stage

• Automate text and sentiment analysisUsing natural language processing (NLP) technology, AI can automatically extract topics, quote keywords, judge emotional tendencies (positive, negative, neutral) and identify intents on a large number of text data such as user comments, interview records, and open-ended questionnaire answers.
• User behavior pattern miningThe AI platform can collect and analyze user behavior data such as clickflow, browsing path, and dwell time within the product in real time, and automatically identify typical user behavior patterns, high-frequency usage scenarios, and potential pain points.
• Large-scale data processing capabilities: In the face of massive research data, AI has shown processing speed and capabilities that far exceed manual processing, and can complete data cleaning, aggregation, and preliminary analysis in a short period of time.
• Cross-analysis and association discoveryAI models can uncover hidden associations and deep patterns from multi-dimensional, seemingly unrelated data points, uncovering insights that are difficult for humans to detect.

3.4 Report generation and insight extraction stage

• AI automatically generates the first draft of the insight reportBased on the analysis results, AI can automatically integrate key findings, distill core insights, and generate a structured first draft of the research report, even including preliminary conclusions and recommendations.
• Data visualization assistanceAI tools often have powerful built-in data visualization engines that can quickly transform analysis results into easy-to-understand charts, dashboards, and more.
• Predictive insights: Through machine learning models, AI can predict future user behavior trends, market changes, or potential needs based on existing data, providing prospective support for product decision-making.
• Personalized report customization: AI can adjust the focus, presentation, and level of detail of reports according to the needs of different audiences (such as management, product teams, marketing teams).

4. In-depth analysis of core AI user research tools

There are many AI tools available in the market that can be applied to different stages of user research. The following will be a review of Flowith, Whale, TYPICA. AI and Graphy, four representative tools, for in-depth analysis.

4.1 Flowith: Knowledge-driven AI creation and research assistant

Core functions and features：

• Visualize the knowledge garden: Flowith can automatically parse various materials (documents, notes, web links, etc.) uploaded by users into structured “knowledge seeds” and build a dynamic and scalable knowledge network graph. This makes information precipitation, semantic retrieval, and associative calls more convenient.
• Oracle Task Autonomy Mode: This is a feature of Flowith. When a user proposes a complex task (such as “analyze this user interview summary and generate a report”), Oracle Mode can independently plan and break it down into multiple sub-steps, and then automatically select and call appropriate AI tools (such as content summarization, mind map generation, PPT production, web page generation, etc.) to complete it, without the need for users to write complex prompts.
• Multi-model and multi-threaded support: Users can parallel scheduling and using multiple different AI large models (such as GPT series, Claude, etc.) to collaborate on tasks within the same interactive Canvas interface, making full use of the advantages of each model.
• Team collaboration: Flowith supports multi-person real-time collaboration across knowledge bases, canvases, and workflows, allowing team members to jointly manage knowledge assets, conduct research and analysis, and create content.

User research application scenarios and potential：

• In-depth integration and analysis of research data: A large number of user interview recordings can be transcribed into transcripts, questionnaire answers, second-hand research reports, etc. can be imported into Flowith’s knowledge garden to form a structured research database. Utilize its semantic understanding and association capabilities to conduct in-depth topic mining, user pain point aggregation, and demand pattern recognition.
• Automate the generation of first draft reports: Through Oracle mode, set the framework and requirements of the research report, Flowith can automatically extract relevant analysis results, user sounds, data charts, etc. from the knowledge garden, and quickly generate a structured first draft of the research report, PPT presentation, or interactive web page summary.
• Exploratory research on complex research topics: For some exploratory research topics that are not clearly defined and require multiple rounds of information gathering and iterative analysis, Flowith’s multi-step disassembly and multi-model collaboration capabilities can provide strong support.
• Work as a team to complete large-scale research projects: The research team can deposit all relevant data and analysis processes in a shared Flowith workspace for efficient collaboration, knowledge sharing, and unified results.

advantage：

• High degree of automation: Oracle model autonomously plans and executes complex tasks, greatly reducing manual intervention.
• Strong knowledge precipitation and reuse ability: The knowledge garden enables the effective accumulation and convenient reuse of research materials and insights, avoiding information silos.
• Handle massive amounts of complex information: Especially good at working with unstructured text data and extracting value from it.
• Support team collaboration: Optimize team research processes and knowledge management.

limitations：

• Not a full-process research tool: Flowith focuses more on data processing, analysis and content generation in the middle and late stages of research, and lacks front-end research execution functions such as user recruitment, questionnaire design and distribution.
• Learning curve: Its powerful features and unique interaction patterns may require users to learn and adapt to it.
• Network access restrictions: Users in some regions may have network stability issues when accessing overseas AI services.
• Result accuracy dependent: The accuracy and depth of AI-generated content still need to be reviewed and optimized manually.

4.2 Yujing: Chinese information processing and intelligent reading expert

Core functions and features：

• Strong Chinese semantic understanding and text generation capabilities: Based on the large model of Shenyan Technology, Yujing has an industry-leading level in Chinese processing and better understands Chinese context and expression habits.
• Intelligent text information processing: Core features include document processing, RSS feeding, note management, and intelligent Q&A based on document content. It can help users quickly understand and process massive amounts of information.
• Information aggregation and subscription management (especially good at WeChat official accounts): Yujing can effectively aggregate information sources from different channels, especially in aggregating the content of WeChat official accounts, which can help users break down information barriers.
• AI-assisted reading and note management: Automatically generate article guides, summaries, and outlines, and support asking questions about article content and getting answers based on the original text. Its note excerpts and management functions are also quite distinctive.

User research application scenarios and potential：

• Preliminary desktop research and dynamic tracking of competitors: Before the start of user research, researchers can use Yujing to efficiently aggregate, read, and summarize background information such as industry reports, research papers, news information, and official account articles published by competitors related to the research topic, and quickly establish a knowledge reserve.
• Preliminary processing of user interviews and open question texts: Import a large number of open-ended responses from interview transcripts or questionnaires into Yujing, and use its AI summarization, outline generation, and intelligent Q&A functions to quickly grasp the core content, extract key information points, and assist in subsequent in-depth coding and analysis.
• Assist in the collation and writing of research report materials: In the report writing stage, Yujing’s note management and text generation capabilities can help researchers organize the materials in the analysis stage and provide structure and content reference for certain chapters of the report (such as background introduction and literature review).

advantage：

• Excellent Chinese processing ability: Excellent in understanding and generating Chinese content, ideal for handling Chinese research materials.
• Efficient information acquisition and digestion: Powerful information aggregation (especially WeChat official accounts) and AI-assisted reading functions can significantly improve the efficiency of researchers in processing massive amounts of text information.
• User-friendly note-taking system: Convenient for researchers to record, organize and review inspiration and insights at any time during the reading and analysis process.

limitations：

• Non-dedicated user research tools: The core positioning of Yujing is intelligent reading and information processing assistant, and does not have special functions for user research such as user recruitment, questionnaire design, and data statistical analysis.
• Data analysis ability is basic: Its AI analysis functions mainly focus on text summarization and information extraction, and cannot replace professional qualitative/quantitative data analysis methods and tools.
• Focus on personal use scenarios: At present, the product form is more inclined to personal information management and knowledge improvement, and the support for teamwork research scenarios is relatively limited.

4.3 TYPICA. AI: AI research companies focusing on language, culture, and context

Core functions and features：

• Deep localization and cultural adaptabilityTYPICA.AI’s core advantage lies in its AI models being trained on local languages (such as Arabic and its dialects Darija, French, etc.) and cultural backgrounds, emphasizing a deep understanding of nuanced context and cultural connotations.
• AI capabilities for complex contexts: Provides AI capabilities such as information extraction, semantic search, and multilingual understanding, and is designed for localization environments and low-resource language scenarios where “conventional AI is not competent”.
• System transparency and interpretability: Committed to building explainable AI systems, rejecting the “black box” of technology, and emphasizing application transparency and traceability.
• Reach vulnerable languages: Focus on and support non-major languages and dialects that mainstream AI technologies often overlook.

User research application scenarios and potential：

• Data analysis from multilingual and cross-cultural user surveys: When user research involves user groups in non-English or other mainstream languages, especially interviews and social media texts that contain a large number of local dialects, slang, or culturally specific expressions, TYPICA. AI can provide more accurate semantic understanding and sentiment analysis.
• Deep insights for users in specific regional markets: When conducting product localization or market entry studies for specific cultural regions (e.g., the Middle East, North Africa, etc.), TYPICA. AI helps to more accurately grasp the real needs, preferences, pain points, and cultural taboos of local users.
• Handle colloquial, non-standardized user feedback: For natural, colloquial and even misspelled text feedback generated by users in social media, customer service conversations, etc., TYPICA. AI’s contextual understanding capabilities may be more advantageous.

advantage：

• Strong localization adaptability: Better understanding of region-specific languages, dialects, and subcultures than generic AI tools, improving the quality of insights extracted from user data.
• High accuracy of contextual understanding: Better handles culturally relevant nuances and implicit meanings.
• Cover the blind spot of general AI: Provides a feasible AI solution for user research in vulnerable languages and specific cultural contexts.
• The system is highly transparent: Helps build trust in AI analysis results, especially in innovative research projects that require compliance and traceability.

limitations：

• It is not a full-process user research platform:TYPICA.AI mainly focuses on language technology such as natural language understanding and information extraction, and the support for standard research processes such as user recruitment, questionnaire generation/distribution, and report automation is still unclear.
• There are few public user research cases: At present, there are not many official disclosed complete user research application cases, and its ease of use, efficiency improvement and depth of insight in real research projects need to be verified by more practice.
• Functional universality may be limited: Its core advantage lies in specific language and cultural scenarios, and may not be the first choice for general, standardized, and large-scale user research projects.
• Rely on on-premises data resources: The model’s performance may be limited by the quantity and quality of training data available in the target language or region.

4.4 Graphy: AI-powered data visualization and report generator

Core functions and features：

• AI intelligently generates charts: Users can use simple natural language text descriptions (e.g., “show user growth by channel over the past year”) or upload data, and Graphy can intelligently identify intent and automatically generate corresponding charts.
• Diverse chart types and rich customization options: Supports various common chart types such as bar charts, line charts, pie charts, scatter charts, funnel charts, maps, etc., and provides flexible customization functions such as color, font, label, legend, background, and layout.
• Convenient data import: You can directly import data files in Excel, CSV, Word and other formats, and also support connecting data sources from online tabular tools such as Google Sheets.
• AI-Generated Data Insights and Chart Text Explanations (Beta): Graphy attempts to use AI to analyze chart data and automatically generate text interpretations and summaries of trends, key indicators, anomalies, etc., to assist users in understanding the data.
• Chart export and sharing: Supports exporting charts to various formats such as images (PNG), PDF, HTML, etc., which are convenient for embedding into reports, presentations, or web pages.

User research application scenarios and potential：

• Fast visualization of questionnaire quantitative data: Import the statistical data collected from the questionnaire (such as the proportion of multiple choice questions, the average value of scoring questions, etc.) into Graphy to quickly generate intuitive and beautiful charts for preliminary analysis and result display.
• Data presentation and beautification of the research report: When writing user research reports, charts made with Graphy can make the data more expressive and persuasive, improving the professionalism and readability of the report.
• Assists in data analysis and insight extraction: Use Graphy’s AI automatic text description function (although it is currently mainly supported in English and in beta), you can get a preliminary interpretation of the data graph and inspire analysis ideas.
• Dynamic dashboards display the core indicators of the survey: If your survey involves tracking certain user metrics on an ongoing basis, you can use Graphy to create updatable dashboards.

advantage：

• Ease of use and low threshold: Quickly generate professional charts with simple text descriptions or data uploads without the need for professional diagramming or programming skills.
• Chart generation is efficient and saves time: Automates the chart creation process, significantly reducing the time spent manually adjusting charts in tools like Excel.
• Enhance the appreciation and professionalism of the report: The generated charts are rich and beautiful, which helps to improve the display effect of research results.
• Support multiple data import methods: Facilitate the integration of survey data from different channels.

limitations：

• It is not a full-process user research tool: The core function of Graphy is data visualization, which is mainly used in the later data analysis and report writing stage of user research, and does not involve research design, execution and other links.
• AI analysis capabilities are not yet mature: Its AI-generated data insights and text descriptions are still in its early stages, with limited accuracy and depth, and mainly support English and insufficient support for Chinese.
• Limited advanced customization capabilities: While it offers basic customization options, it may not be a complete replacement for professional BI tools for extremely complex chart customization needs or in-depth interactive data exploration.
• Data security considerations for online tools: For survey data containing sensitive business information or user privacy, it is necessary to carefully evaluate its data security and privacy protection policies when uploading it to online platforms.