‍

AI is already helping researchers write questions, summarize interviews, and identify themes. But the bigger opportunity is what happens when AI moves beyond individual tasks and starts helping coordinate the entire research workflow.

In our latest Hubble webinar, we demonstrated how an AI agent can work with Hubble through MCP to run an end-to-end user research study — from planning and study creation to recruitment, moderation, analysis, and synthesis. Instead of moving between different tools and manually carrying context from one step to another, the researcher can manage much of the workflow through a single conversation.

What Is Agentic User Research?

Most applications of AI in user research today focus on a specific task. You might use AI to generate a discussion guide, summarize an interview transcript, or identify themes across participant responses. Each of these can save researchers time, but the researcher still needs to coordinate the overall process.

Agentic user research takes this a step further. By connecting an AI agent to research tools and data through MCP (Model Context Protocol), the agent can take actions on behalf of the researcher. It can retrieve previous research, create studies, launch new research, analyze results, and use what it learns at one stage to inform what happens next.

The researcher remains responsible for the important decisions. The agent helps execute the workflow around those decisions.

Planning Research with Existing Context

A research project often begins before the first question is written. Researchers need to understand what the team already knows, identify gaps in existing research, determine which users they need to speak with, and decide which methodology makes the most sense. With access to a research repository through MCP, an AI agent can use previous research as context when helping plan a new study.

Instead of starting from a blank page, researchers can ask questions such as:

• What have we already learned about this problem?
• What questions remain unanswered?
• Which user segments should we investigate?
• What methodology would help us answer these questions?

This turns the research repository into active context for planning new research.

Creating and Launching the Study

Once the research plan is established, the agent can help turn that plan into an actual study. Through Hubble's MCP, the agent can create the study structure, generate questions and tasks, configure the research, and prepare it for participants. The researcher can review and refine the study before launching it, while the agent handles much of the repetitive setup work. This creates a tighter connection between research planning and research execution.

Conducting Research with AI

The same workflow can extend into data collection. For qualitative research, AI-moderated interviews can conduct conversations with participants and dynamically probe their responses. For usability research, participants can complete tasks while researchers capture behavioral and qualitative feedback. Rather than treating AI moderation as a separate workflow, an agent can incorporate it as one part of a larger research process. The study that was planned in the conversation can become the study participants actually experience.

Analyzing and Synthesizing the Results

Once responses begin coming in, the agent can work with the resulting research data. Researchers can ask the agent to identify recurring themes, compare responses across participants, surface important quotes or observations, and summarize the most important findings. Because the agent has context from the original research plan, analysis doesn't have to happen in isolation. It can connect the findings back to the original questions the team was trying to answer.

For example: Research question → Study → Participant feedback → Themes → Insights

Instead of manually moving information through each stage, context can carry across the entire workflow.

The Researcher Is Still in Control

Agentic research doesn't mean handing research decisions over to AI. Researchers still determine what questions are worth investigating, which participants matter, whether the methodology is appropriate, and whether the resulting evidence actually supports a conclusion. The difference is where researchers spend their time.

Rather than spending as much time configuring tools, moving data between systems, organizing responses, and repeating administrative work, researchers can spend more time interpreting evidence and making decisions. AI agents handle more of the execution layer around the research.

The Key Takeaway

The most interesting opportunity for AI in user research may not be automating individual research tasks; it's connecting them. When AI agents can securely interact with research tools and repositories, planning, execution, recruitment, moderation, and analysis can become parts of one continuous workflow.

Researchers remain responsible for judgment and research quality, while agents help carry context and execute the repetitive work between each stage. In our latest webinar, we walk through this workflow end-to-end and demonstrate what an agentic research process looks like in practice.

‍