Changelog

Check now runs in under 5 minutes!

We’ve rebuilt core parts of Check to dramatically reduce total runtime.

From the moment you upload your dataset to the moment you export flagged respondents, the full process now completes in under 10 minutes.

Alongside the speed improvements, we’ve made Check clearer and more flexible:

More control over flags

Manually select and unselect respondents you don’t want labeled as outliers.

Clearer live previews

Live previews now only show columns relevant to each check — including Open-Ended checks, Response Speed, and Chuck.

Smarter time detection

Check now supports both total survey time columns and Start/End time columns, automatically calculating response time either way.

Less waiting. Cleaner reviews. More control.

Upload more. Boost faster.

Previously, uploading more than 20 boosts at once could overload the system and lead to failures.

That limit is gone.

Boosts now run in parallel, so you can upload and process as many files as you need — all at once, without interruptions.

No more workspaces

Say goodbye to workspaces.

You no longer need to create workspaces or assign credits.

Just sign in and see all of your projects in one place.

Fewer steps. A simpler way to work.

LLM Quality Inspector Agent (Chuck)

We’ve added Agentic Checks, powered by our new LLM inspector, Chuck.

Chuck identifies contradictions in your survey that typically only a human reviewer would catch. He does this by analyzing your survey structure and automatically grouping questions that should be evaluated together — including funnels, opposites, repeated brand sets, and behavioral sequences. For each group, Chuck:

• Provides a short explanation of why the questions were grouped
• Allows you to review or edit the grouping
• Check each respondent for contradictions or logically inconsistent answer patterns

This update adds deeper context checks with no manual configuration.

Updates to Fairgen Check

Based on Beta feedback, we’ve released several workflow and clarity improvements:

• XLSX downloads for quick offline analysis
• Manual check/uncheck controls for adjusting flagged respondent selections
• Improved live preview tables that highlight anomalies more clearly
• Shorter, clearer flag explanations to make decisions easier

REST API for Boost Generation

You can now run Fairgen Boost fully through our REST API.

The API supports:

• Dataset upload
• Structure upload
• Niche selection
• End-to-end Boost generation

This allows automated or large-scale workflows without using the UI.

Decipher XML Support

Fairgen now accepts XML survey structure uploads.

Our autodetection algorithm reads the XML file and automatically identifies derived variables/ recodings. This removes the need for manual recode setup and reduces opportunities for configuration errors.

Upload the XML and Fairgen handles structure interpretation automatically.