The Rose platform
Open what your company already learned.
Rose is the AI research partner for insights teams. It opens what your company already learned, for people and agents, with the source attached, and links findings across studies. When the answer doesn’t exist yet, Rose helps scope and orchestrate the new research, including a draft the team reviews.
LIBRARY · 8 STUDIES
- S1Household panelSurvey · 2,000 respondents
- S2Churn interviewsQual · 24 interviews
- S3Cancellation surveySurvey · exit flow
- S4Segmentation studyQuant · segments
- S5Dinner diary studyQual · video diaries
- S6Food waste deep diveMixed methods
ASK ROSE
Where do the survey and the interviews disagree?
On price. The cancellation survey ranks price first S3, but in interviews price comes up only after variety and waste S2. Price is likely the easiest reason to give, not the main one.
DELIVERABLE
REPORT · DRAFT
Early cancellations: what the evidence says
Price is the reason people give, not the reason they leave. Variety and waste come first in interviews S2, while the exit survey ranks price first S3.
Rose Knows
What your company already learned, with the source attached.
Open the research people and agents are asking for.
People and agents can ask what the company already learned, across reports, trackers, survey datasets, transcripts and presentations.
The source stays attached, so a reader can check the passage behind each part of the answer.
EXAMPLE
- What drives customers to switch, and how has that changed?
- Which frustrations show up in both the store research and the digital research?
- How have audience expectations shifted across the studies you already ran?
Stop being the research help desk.
Why do subscribers cancel in the first 90 days?
Three causes recur. Weeks two to four feel repetitive once the novelty fades S2S5. Skipping a week is harder than cancelling S3. And households of one see too much waste S1S6.
SOURCE PASSAGES
By week three it felt like the same five dinners.S2 · Churn interviews · participant 14
I only wanted to pause. Cancelling was quicker.S3 · Cancellation survey · open-ended response
Read the survey data behind the study.
Work with the survey data a study already holds: metrics, cross-tabs, wave comparisons, MaxDiff, conjoint, open-end coding and charts.
Where two waves are not comparable, or a figure is not in the source, the answer says so.
EXAMPLE
- Rank what matters most in a MaxDiff study.
- Read the trade-offs in a conjoint study by segment.
- Compare tracker waves, and say when the method changed.
- Group open-ended answers into themes.
The number stays tied to the study it came from.
What matters most in a meal kit
MaxDiff · S1 Household panel · 2,000 respondents
- Finding
- Variety and easy skipping outrank price per meal, in line with the interviews.
- Measure
- Share of preference. Top five of twelve items shown.
Remember how your team talks about the work.
Rose keeps terminology, priorities and research preferences, so the team stops explaining the same things.
That memory supports access. It is not a substitute for the source.
EXAMPLE
- Keep a segment definition the team has already agreed.
- Retain the KPIs a brand-health read is supposed to use.
- Use the product names the team actually says.
Less repeating yourself. The source still has to be there.
- Term
- Active customer
Ordered at least once in the last eight weeks. - Priority
- Early retention is this year’s focus for the insights team.
- Preference
- Break results out by household size before anything else.
- You
- Prefers a one-paragraph summary ahead of the detail.
Rose Connects
The full picture, not one stray slide.
Link findings across studies, teams and years.
Findings from separate studies show up as one picture, with the source attached to each part.
Rose has its own research repository and works alongside the repositories and tools your insights team already uses.
EXAMPLE
- Pull a topic across the studies already in the library.
- See where two teams described the same behavior in different words.
- Keep a link back to the original deck, dataset or transcript.
One picture of what the company already learned.
- S1Household panelSurvey · 2,000 respondents
- S2Churn interviewsQual · 24 interviews
- S3Cancellation surveySurvey · exit flow
- S4Segmentation studyQuant · segments
- S5Dinner diary studyQual · video diaries
- S6Food waste deep diveMixed methods
Carry a method from one study to the next.
A reusable research skill holds the inputs, steps and rules for a kind of work, such as a brand-health read of awareness, consideration and preference.
The skill can require the team’s comparison rules, including when a difference is strong enough to report.
EXAMPLE
- Reuse a brand-health read instead of restating it.
- Apply the same segment definitions in every market.
- Keep the reporting structure the team already trusts.
The same method, the next time the question comes up.
Brand-health analysis
Organization skill · written by your team
- 1Use the tracked KPIs: awareness, consideration, preference.
- 2Compare with the previous wave and the category average.
- 3Flag differences at 95% confidence only.
- 4Report in the quarterly brand-health structure.
Sound like your team.
Structure, tone, terminology and chart choices follow a previous deliverable the team treats as the standard.
EXAMPLE
- Headline, then evidence, then implication.
- The team’s own terms, not a generic glossary.
- A short summary before the detail.
Less reformatting. The evidence still has to be in the draft.
- Structure
- Headline, evidence, implication.
- Tone
- Plain and direct. No jargon.
- Terms
- “Subscribers”, never “users”.
- Charts
- Horizontal bars, largest first.
- Depth
- Two pages, with an appendix for method.
Keep a living view of a topic.
Rose maintains a shared account of what the company knows about a topic, such as retention or brand health, refreshed as relevant studies are added.
The team’s edits stay. A new study updates the picture. It does not start it over.
EXAMPLE
- Hold what is known about early retention in one place.
- Notice when a new study changes the ranking of reasons.
- Keep the sources on the summary as the library grows.
The full picture stays current as new studies arrive.
Early retention
Shared with the insights team · 5 studies
Cancellations cluster in weeks two to four and are driven by repetition and waste more than price S2S3S6.
WHAT CHANGED
- Usage analysis addedConfirms customizers stay longer. Theme refreshed.
- Food waste deep dive addedWaste now ranks second among reasons to leave.
- Theme createdFrom the churn interviews and the cancellation survey.
Rose Acts
When the answer doesn’t exist yet.
Rose helps orchestrate new research. Your team stays in charge of the work that matters.
Scope the research that still needs doing.
When the library cannot answer, Rose helps name what is already known and draft the questions a new study would have to ask.
Your team decides whether that work goes ahead.
EXAMPLE
- A draft brief that starts from what the studies already show.
- Objectives, questions and the gap, in one place.
- The sources behind “what we already know.”
New research gets scoped. Your team decides.
BRIEF · DRAFT
Retention study: weeks two to four
OBJECTIVE
Identify which changes in weeks two to four would cut early cancellations.
RESEARCH QUESTIONS
- Does more menu variety change intent to stay?
- Would easier skipping replace cancelling?
- Do single households want a smaller box?
WHAT WE ALREADY KNOW
A plan the team can edit.
Rose lays out the steps, including what to read in the library and what is still missing, and waits for the team to approve or edit the plan.
EXAMPLE
“Why is purchase consideration falling despite improving satisfaction?”
- Compare what the trackers already say.
- Mark the interviews that still need reading.
- Stop until someone on the team approves the plan.
Nothing moves ahead without the team.
Why is purchase consideration falling despite improving satisfaction?
- ✓Compare consideration and satisfaction across tracker waves.
- ✓Break both out by segment and household size.
- ●Read the interviews for reasons people stopped considering.
- ○Draft findings with charts and limitations.
A draft the team reviews.
Rose drafts an account of the evidence, with the source attached, that a researcher can revise before anyone else sees it.
EXAMPLE
- A draft report the team can revise section by section.
- Charts that stay tied to the study they came from.
- A clear place for what the evidence does not support.
The team reviews the draft. Rose does not send it.
REPORT · DRAFT
Early cancellations: what the evidence says
Price is the reason people give, not the reason they leave. Variety and waste come first in interviews S2, while the exit survey ranks price first S3.
RECOMMENDATIONS
- Rotate menus in weeks two to four.
- Make skipping a week a one-step action.
- Offer a smaller box for households of one.
What Rose is not
Clear about what it does. And what it doesn’t.
Not a survey tool.
Rose doesn’t field surveys or recruit participants. It works with the research you already have. When that cannot answer a question, Rose helps scope the next study, and your team stays in charge.
Not a general-purpose chatbot.
Rose answers from your organization’s research, with the source attached, not from the open web. A reader can check the evidence.
Not a replacement for researchers.
Researchers review the plan, refine the draft and decide what goes out. Rose supports your team’s judgment. It doesn’t stand in for it.
Trust & security
Enterprise research deserves enterprise-grade safeguards.
Rose helps teams use AI with confidential research while maintaining control over access, evidence and deliverables.
Protect your organization's knowledge.
Rose uses organization-level data isolation and access controls to separate research across customers and teams.
Control access and sharing.
Manage access to research and deliverables, with support for organization-level identity and user-management capabilities.
Inspect the evidence.
Rose links factual claims to supporting research and runs checks on citations and calculations. Researchers can examine the underlying evidence instead of accepting unsupported conclusions.
Keep researchers in control.
Review research plans, refine reports, and approve important deliverables. Rose supports your team's judgment rather than replacing it.
Security and research rigor aren’t optional features. They’re fundamental to how Rose works.
Learn About Security & AccuracyQuestions
Product questions.
Rose searches across reports, trackers, survey datasets, transcripts and presentations, and keeps the source attached to what the company already learned.
Yes. Rose works with the survey data a study already holds, including metrics, cross-tabs, wave comparisons, difference tests, open-end coding and charts, and says when the evidence limits the conclusion.
Rose links factual claims to the supporting research and runs checks on citations and calculations, so a researcher can inspect the evidence.
Yes. Rose applies reusable research skills, including organization-specific skills, that hold how a kind of work should be done, and uses previous deliverables as examples of structure, tone and terminology.
Rose produces cited answers with the source attached and links across studies. It also helps scope and orchestrate new research when the answer doesn’t exist yet, including a draft the team reviews before it goes anywhere.
No. Rose works with the research you already have. When that cannot answer a question, Rose helps scope and orchestrate the new research. The team stays in charge.
Give them access to what you already know.
Open what your company already learned, for people and agents, with the source attached. Rose helps orchestrate the research that still needs doing, with your team in charge.