# Supercharge your AI agent with human feedback

Run user tests, evals, and more—from the tools you already use.

## How it works

### Your agent calls Voicepanel like any other tool

MCP (Model Context Protocol) lets your assistant call Voicepanel as a built-in tool—studies, audiences, recruiting, voice and video responses, and analysis—without leaving the thread. You steer in plain language; we run the research and return structured results.

## Capabilities

### Everything your agent needs to execute user research

From project design to final insights, every step is accessible programmatically through a single MCP connection.

01  
### Create & configure projects  
Your agent defines the research objective, writes the questions, and sets the parameters. Open-ended, multiple choice, video—whatever the research calls for.

02  
### Define target audiences  
Specify demographics, behaviors, or custom screening criteria. Recruit from Voicepanel's integrated panel of 30M+ participants or send to your own customers.

03  
### Collect voice, video, and text responses  
Real people respond in their own words through voice or video. Voicepanel's AI moderator probes deeper based on what they say, just like a live interview.

04  
### Run AI-powered analysis  
Responses are automatically transcribed, themed, and analyzed. Your agent gets back structured insights: segments, sentiment, key themes, and supporting quotes.

05  
### Iterate in real time  
Your agent can check project progress, review early results, and adjust course. Launch follow-up projects based on what it learns. The feedback loop is continuous.

## Steer product development with human feedback at every step

Build with confidence: every decision your agent makes can be informed by what real people think, feel, and prefer.

### AI prototyping  
Validate generated prototypes with real users before you commit serious engineering time. Your agent can iterate on concepts early so you double down on directions people actually want.

### Campaign development  
Test AI-generated marketing concepts, creatives, and messaging with multiple target audiences in minutes. Launch campaigns with confidence knowing how your target segments will react.

### AI evaluations  
Recruit evaluators, compare model outputs side by side, and collect rankings plus rationale at scale. Feed cleaner human judgments into benchmarks, safety reviews, and tuning workflows.

### User segmentation  
Group people by attitudes, needs, and behaviors using real feedback—not only demographics. Your agent can run studies that feed richer segments into targeting, messaging, and roadmaps.

### Continuous discovery  
Run lightweight studies whenever a new question comes up, instead of waiting for the next big research cycle. Your agent keeps a running read on customers that gets sharper over time.

## The shift

### From building blind to building with conviction

Most AI agents operate in a vacuum. Voicepanel gives them the one thing they're missing: direct access to the people they're building for.

### Today

—Hundreds of agents shipping features nobody asked for  
—Research happens quarterly, months after decisions are made  
—Teams guess what users want, then build it anyway  
—Teams spend hours context switching between different tools  
—No shared understanding of the customer across teams

### Future

→ Focused agents that validate every idea with real humans first  
→ Continuous feedback loops running alongside every sprint  
→ Every product bet is backed by hundreds of voice responses  
→ Research and shipping share the same agent workflow  
→ A living, evolving picture of your customer shared across every agent

## MCP FAQ

### Still stuck? Contact us to request access.

**When is Voicepanel MCP available?**  
Our MCP server is available now. We're onboarding teams in waves so we can support each integration properly. Request access from this page—we'll reply with connection details and next steps.

**What can I do with Voicepanel's MCP?**  
We've worked hard to expose all of Voicepanel through MCP—projects, distributions, responses, analysis, and more. Your agent only sees what your account role already allows; nothing bypasses in-app permissions.

**Which AI tools and clients can I use?**  
Any assistant, IDE, or automation stack that supports custom MCP servers over HTTP with OAuth can connect—think Claude, ChatGPT where MCP is enabled, Cursor, Windsurf, and open-source MCP clients. If your tool lets you add a server URL and complete an OAuth login to Voicepanel, you're set.

**What tools does the MCP server expose?**  
The server exposes many small, purpose-built tools. They fall into two groups:

- **Read tools:** Account context, studies, plans, distributions, response lists and transcripts, study stats, analysis questions and insights, segmentations, and more. Everyone who can sign in to Voicepanel with a valid role gets these— including viewers—so agents can safely explore and summarize data within their permitted scope.
- **Write tools:** Create or generate studies, edit studies, create and edit distributions, publish panel recruits, update responses and distribution status, run analysis, request study review, create segmentations, and related actions. These are available to editors and admins only; viewers stay read-only. The exact tool list your client sees is filtered per user.

**Why did you launch MCP?**  
Teams already work inside AI tools—we wanted Voicepanel there too, with the same data and permissions as the product. MCP is an open standard, so one connector works across assistants instead of a custom integration every time.

**How is MCP priced?**  
There is no separate MCP line item. Connector access is included with your Voicepanel subscription at no additional cost—same account, same roles, same data limits you're already on.