Chat is the AI assistant built into the MX8 Labs Research Platform. It works with whatever you already have open, whether that is a survey, a report, an insight, a checklist, or a data-prep script. Because it runs inside the platform, Chat can read the current project and survey context directly and propose specific changes. Survey-code and data-preparation edits are proposed for review: Apply updates the editor draft, and Save persists it. Creating or configuring reports and insights takes effect when the action runs and can be edited or removed afterwards.
The sections below cover what Chat can do across the survey lifecycle. For more examples, see the Chat example actions page.
Review survey and data-preparation changes
Chat proposes; you decide. The assistant can draft new wording, rewrite survey code, adjust logic, or clear checklist issues, but it does not save those changes to your survey by itself. It shows a comparison of the current draft and proposed code first. Reject leaves your draft unchanged. Apply puts the proposal into the editor draft; review and select Save to persist it.
A questionnaire is the foundation of a study, and a single unreviewed change to wording, logic, or quotas can bias results or break fielding. Chat is designed to keep that review step explicit. You can ask for a large change, check the diff, and accept or discard it without touching the live survey.
The same review workflow applies to survey-code and data-preparation recode proposals. Use the comparison controls to move through changes before applying them. If the editor draft changes while a proposal is pending, reject it and request a fresh proposal against the current draft. Report and insight actions still take effect as they run; inspect and adjust them in their own editors.
Working in context, not on pasted text
Chat works with the live object you have open rather than a screenshot or a block of copied text. It can see the active survey or project, the current tab, the reports and insights that already exist, the go-live checklist, the media table, translation status, and quota context. That means you can ask it to summarize where a project stands, hand work off to a teammate, or pick up where you left off, and it will answer from the actual state of your work.
Chat helps you work with results; it does not replace the platform's analytical engine. Cross-tabulation, weighting, and statistical testing are still performed by MX8 Labs, with significance and thresholds preserved as configured. Chat helps you set up, interpret, and act on those results rather than recalculating them.
Finding and navigating your workspace
Chat is also the fastest way to move around the platform. Ask it to find something by name or keyword and it runs the same universal search as the search bar in the top app bar, covering projects, surveys, reports, and insights across your workspace — not just the item you have open. From what it finds, Chat can take you directly to the right screen: it can open a project, survey, report, insight, or account, and even land on a specific tab, such as a survey's Reports, Quotas, Media, or Translations tab. That turns a request like "find the Q2 brand tracker and open its reports" into a single step instead of a hunt through the project list. Navigation only moves you around the app — it changes nothing in your work.
Reviewing and improving surveys
Chat can review a draft survey and suggest improvements the way an experienced research lead would. It looks for leading or loaded wording, double-barreled questions, ambiguous or overly complex phrasing, weak response options, inconsistent scales, and opportunities to reduce order bias through randomization. It also checks that the questionnaire flows sensibly, that terminology and formatting stay consistent, and that each question earns its place against the study objectives. This builds on the same principles in our survey editing guide, with the advantage that Chat is reading your actual questionnaire rather than working from generic advice.
If you prefer to work visually, Chat can review the structure shown in the visual survey editor and suggest a cleaner question order, clearer section labels and comments, simpler conditional logic, and better placement of consent and termination steps, all with an eye to respondent burden and readability.
Reading and updating survey code
When you have a survey open, Chat can inspect the survey program and explain what the code is doing in plain English. It can identify broken or risky logic, rewrite sections more cleanly, add or adjust skip logic and validation, refine quotas and branching, and insert calculated variables. Because the platform offers full parity between the visual editor and code, you can ask Chat to make the same change either way. Use Edit with Chat to request a change from the editor. Chat proposes an update against the current draft; review it, select Apply, then Save.
Chat can also strengthen quality controls. It can propose and place attention checks, custom validators, and termination rules, and suggest logic for handling repeated low-quality responses. Our guidance on gotcha questions and broader fraud-prevention measures describes the techniques Chat draws on when recommending safeguards across device, network, and behavioral signals.
Resolving checklist issues before launch
Every survey runs through an automated go-live checklist as part of testing and review. Chat can read that checklist for the active survey, explain each issue in plain language, and help you prioritize what matters most before launch. From there it can draft the exact survey updates needed to clear each item, in code or the visual editor, and stage them for your review. Review the proposal, select Apply to update the draft, then Save to persist it.
Translations and multilingual fielding
For multilingual surveys, Chat can review translation status for the active survey, flag missing or incomplete translations, and check wording consistency across languages. It can also look at your source text and point out phrases likely to translate poorly, suggesting edits that reduce ambiguity before you field internationally.
Quotas, media, and field readiness
Chat can summarize the current quota report, flag underfilled cells or setup risks, and explain what a given quota issue means for fielding, often suggesting wording, routing, or source changes that would ease the problem. For studies that depend on stimuli, such as ad, concept, and creative testing, it can inspect the media attached to a survey, confirm what is present, identify missing or inconsistently named assets, and check that the media setup matches the survey logic.
Building and interpreting reports
Chat can create reports in the current survey or project and configure them for the analysis you need, including cross-tab, time-series, lift, weighting, and example-verbatim reports. It can select the right questions, set confidence levels and minimum respondent thresholds, and update an existing report to include new tracking questions, all within the standard reporting workflow.
It is equally useful once a report exists. Chat can explain what a report includes, which questions are selected, and whether the chosen analysis settings make sense, including whether the statistical test or minimum sample threshold is appropriate. When you are preparing for a stakeholder readout, it can recommend which questions to feature for an executive summary and help you interpret the significant differences the platform has flagged.
Creating and refining insights
Chat can create Insights for the active survey or project from a research question you provide, and help you sharpen the brief so the output stays focused. Because insights in MX8 Labs answer specific questions using report data and statistically significant findings, a well-framed question matters, and Chat can help you draft one, tailor the framing to a particular audience, or compare the latest waves of a tracker at the project level.
Preparing and recoding data
After fielding, Chat can inspect and update a survey's data-preparation recode script. It can clean or regroup response values, harmonize categories across waves, fix coding mistakes, prepare variables for reporting, and review recode errors, explaining what changed so you can keep tracker waves aligned cleanly and reporting consistent. Chat opens recode proposals for review in the Data Prep editor. Check the comparison, select Apply to update the draft, then Save to persist the script. Reject leaves the draft unchanged.
Getting help inside the platform
Wherever you are in a project, you can ask Chat to review your survey, resolve QA issues, build a report, refine an insight, or update survey logic directly in the app. For a longer list of prompts to get started, see the Chat example actions page.