MX8 Labs Assist 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, Assist can read the current project and survey context directly and propose specific changes. It never saves changes to your survey on its own; every edit is reviewed and applied by you.
The sections below cover what Assist can do across the survey lifecycle. For more examples, see the MX8 Labs Assist example actions page.
Every change is human supervised
Assist proposes; you decide. The assistant can draft new wording, rewrite survey code, set up a report, or fix a recode script, but it does not save those changes to your survey by itself. It shows the proposed edit first, and the edit is only written to the survey when you review it and choose to apply it. Until you do, your survey is untouched.
A questionnaire is the foundation of a study, and a single unreviewed change to wording, logic, or quotas can bias results or break fielding. Assist 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.
This principle holds everywhere Assist can take action. Whether it is updating survey code, clearing checklist issues, building a report, or rewriting a data-prep script, the change is staged for your approval first and saved only when you apply it.
Working in context, not on pasted text
Assist 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.
Assist 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. Assist helps you set up, interpret, and act on those results rather than recalculating them.
Reviewing and improving surveys
Assist 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 Assist is reading your actual questionnaire rather than working from generic advice.
If you prefer to work visually, Assist 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, Assist 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 Assist to make the same change either way. As with the existing AI Copilot workflow, Assist shows you the proposed update and writes it to the survey only when you review and apply it.
Assist 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 Assist 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. Assist 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. As with every Assist action, nothing is saved to the survey until you approve and apply it.
Translations and multilingual fielding
For multilingual surveys, Assist 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
Assist 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
Assist 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. Assist 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
Assist 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 Assist 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, Assist 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.
Getting help inside the platform
Wherever you are in a project, you can ask Assist 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 MX8 Labs Assist example actions page.