[Changelog] What's new with the Rally Agent and MCP
Last updated: August 9, 2026
Wednesday July 22nd: Rally Agent + MCP: Beta is Live
The Rally Agent and Rally MCP are now live in Beta, and you can opt in today.
If this is your first time hearing about Rally Agent or Rally MCP, here's what you need to know.
Rally Agent is a new experience inside Rally, built for product builders. Instead of navigating the Rally Research app (which you’re likely pretty familiar with at this point!), your PMs and designers can chat directly with the Rally Agent and build their own studies within the guardrails you set. No training required, no research expertise assumed, no ability to go rogue.
Rally MCP takes that same capability outside Rally. It connects to AI tools your team already uses, like Claude or Cursor, so anyone can run research from inside their existing workflow (and within the guardrails you set) without ever opening Rally.
Two entry points. The same Rally research infrastructure underneath.
Built for research. Grounded in governance.
Rally Agent is built on the governance your research team has already established. Every action automatically follows your engagement rules, cooldown periods, consent requirements, contact limits, and approval workflows, so product teams can move quickly while Research Ops stays in control.
Rally Agent focuses on the operational work that slows research down. It helps teams create studies, recruit participants, schedule interviews, manage incentives, and handle the logistics that turn a research question into a completed session.
What's Rally Agent, and what's Rally MCP?
These are two different ways to run Research in Rally. Same research infrastructure underneath – 2 different entry points.
Rally Agent | Rally MCP | |
Where it lives | Inside the Rally app | In your existing AI tools (Claude, Cursor, etc.) |
Setup required | None | Connect via MCP |
Best for | In-app, conversational research | Teams embedding Rally into custom AI workflows |
Rally Agent is built into your Rally workspace. Open it, type a request, it acts. No configuration required.
Rally MCP is a Model Context Protocol server. It connects Rally to MCP-compatible AI tools so you can drive research from inside Claude, build Rally into a custom agent, or embed it in a team workflow outside Rally's interface entirely.
Get started with the Rally Agent & MCP
Below is everything that's live across both the Agent and MCP, with a small set of MCP-only features noted separately.
❓ Ask questions about your research
Get instant answers without clicking through the app.
"Who's booked this week?"
"Are we on track to hit our participant target?"
"What did P3 say about pricing in their session?"
"Which studies are still in draft?"
"Find customers in San Francisco who've shopped online in the last 30 days."
"What needs my input across my studies?"
⚒ Build and launch a 1:1 interview study
End-to-end study setup, start to finish, in a conversation.
"Set up an onboarding interview study."
"Import this CSV and start a study with these people."
"Add a screener question about company size."
"Drop the age question from the screener."
"Bump the incentive to $75 and change the duration to 45 minutes."
"Attach our standard consent form."
"Publish the onboarding study."
💬 Manage interviews
Run your full interview schedule without leaving your AI tool.
"Reschedule Jane's interview to Thursday at 2pm."
"Cancel tomorrow's session with Alyssa."
"Book Adam for Tuesday at 2pm."
"Move tomorrow's 2pm interview to Sarah."
"Add Jamie as a guest to Friday's session."
"Derek was a no-show, mark it."
"Block my Fridays from the scheduler."
"Prep me for my 3pm with Markus." (generates a full interview brief)
⚙ Handle participant logistics
Make one-off updates to your studies on the fly.
"Add 20 more people to the onboarding study."
"Disqualify Mo from my study."
"Show me Hector's screener responses."
🫡 Stay on top of recruiting
Watch invites and targeting in real time, and adjust the moment something's off.
"Can we target 20 senior engineers in fintech from our existing panel?"
"Approve the recruitment proposal for the onboarding study."
"How many invites went out? Any bounces?"
"Has Adam received my invitation email?"
"Pause recruiting while we fix the screener."
💰 Manage incentives
Approve, adjust, and log payments without leaving the conversation.
"Approve incentive sends for yesterday's interviews."
"Resend Walter's gift card, it bounced."
"Set up a $50 gift card as a custom incentive and log it as sent for Priya."
MCP-only (for now)
A small set of capabilities are available in the MCP but not yet in the in-app Agent:
Create a new population or panel
Look up custom person property definitions
List and preview a study's email campaigns and templates
View detailed consent form status and quota membership data per participant
Workflows to try first
Not sure where to start? Here are some examples to test out first:
Morning standup, AI-style "What interviews do I have today? What needs my input across my studies? Any recruiting decisions pending?"
Build and launch a study in one go "Find customers in our panel who haven't participated in a study. Create an onboarding interview study with a screener that checks for relevant experience. Set the incentive to $75 for 45 minutes."
Interview day, handled "Who am I talking to today? Pull their screener responses and prep me for each session." Then after each session: "Mark P1 as attended. Note that they raised pricing concerns. Approve their incentive."
Recruiting health check "How's the onboarding study tracking? How many invites went out, how many opened, any bounces?"
Reschedule with context "Reschedule my interview with Priya to Thursday at 2pm ET. Show me other available times first."
Coming very soon to Rally MCP
Surveys: Create and manage full survey studies end-to-end.
Unmoderated tests: Set up and launch unmoderated test studies.
External panel recruitment: Recruit through User Interviews and Respondent.
Screener and survey response mapping: Automatically store participant answers in their person profile.
Study collaborators: Add teammates to a study.
The Agent will follow as these capabilities land.
Ready to opt in?
The Beta is free through September 15. Opt in here.
Questions? Reach your CSM or drop us a note at support@rallyuxr.com.
Friday July 31st: Every Study Type, Now in Rally MCP
One week into our Beta and we're already expanding the playbook with top requests 🔥 This week the Rally MCP added full lifecycle support for surveys and unmoderated tests. That means the three most common ways your teams run research are now available through the MCP: interviews, surveys, and unmoderated tests.
Create and launch Surveys via MCP
You can now create, build, launch, manage, and close survey studies end-to-end through the Rally MCP. Write questions in a conversation, publish the study, read responses as they come in, and close it when you hit your target.
What's included:
Create and configure a survey study
Launch the survey
Read responses as they arrive
Pause, resume, or close the study at any time
Use survey integrations (Google Forms, Typeform, etc.) in addition to building questions natively in Rally
Available in: Rally MCP today. Rally Agent coming soon.
Prompts to try this week:
"Create a survey about our new checkout flow with three questions: usability, pricing clarity, and an open text field."
"What did people say on the checkout survey?"
"Pause the checkout survey while we tweak a question."
"Set up a survey for the Onboarding study using this Google Form link: [url]."
Create and launch Unmoderated Tests via MCP
Unmoderated tests are the second most popular study type across Rally workspaces, and they're now fully supported via MCP. Point it at your Maze, UserTesting, Ballpark, or other external testing tool, draft custom messages, and launch. The MCP runs a safety check before go-live and blocks launch if anything critical is missing, so nothing ships half-finished. Monitor completions without logging into Rally Research.
What's included:
Create an unmoderated test with a link to your external tool
Safe go-live validation (blocks launch if link or completion message is missing)
Launch and monitor completions
Pause, resume, or close the test
Available in: Rally MCP today. Rally Agent coming soon.
Prompts to try this week:
“I want to test our new onboarding flow with 10 customers that recently signed up for our free trial. Help me set up an unmoderated test in Maze.”
"How many people have completed the onboarding test?"
More changes that shipped this week
Recruiting approval now happens in the app: When you kick off a recruiting run through the MCP or Rally Agent, review and approval now happen inside Rally instead of inline in Slack or chat. You'll get a link to check exactly who's getting invited and what they'll receive before a single invitation goes out. We did this for another layer of protection so that you can be confident nothing goes out without your explicit permission.
Auto-map screener and survey responses to person profiles: Screener and survey responses can now be automatically mapped to person properties via the MCP. The MCP uses AI to suggest the best property match per question and handles the mapping in one step. Answers get stored on each participant's profile, which means your panel data gets richer with every study that runs.
Coming soon
Automated incentive approvals for surveys and unmoderated tests: same human-in-the-loop approval flow as interviews, coming shortly. For now, log into Rally Research to manually trigger incentives.
Survey and unmoderated test support in Rally Agent: currently MCP only; Agent parity is on the way
New to the Beta?
Rally Agent and MCP are live in Beta and free to test through mid-September. Opt in at the link below and we'll get you set up.
Questions? Reach your CSM or drop us a note at support@rallyuxr.com.