People

The People section is where you manage everyone involved in your research — from real human participants to AI-generated consumer profiles. Build targeted audiences, organize contacts into segments, and track custom attributes to power smarter campaign targeting.

Overview

The People section gives you a complete view of your research audiences. Whether you're building a panel of loyal beauty enthusiasts, creating AI representations of Gen Z consumers, or segmenting your audience by purchase behavior, everything starts here.

What You Can Do in People

• Build AI populations that simulate your target consumers for quick concept testing
• Manage real participants who respond to your video campaigns
• Create segments based on demographics, behavior, or custom criteria
• Track custom attributes like purchase history, preferences, or loyalty tier
• Target the right audience when launching campaigns

Navigation

The People section has five tabs across the top of the page:

Tab	Description
Populations	AI-generated audience profiles representing consumer segments
Humans	Real people who participate in your research campaigns
Molds	Templates for creating consistent persona profiles when mold access is enabled
Lists & Segments	Ways to organize your audiences into reusable groups
Properties	Custom attributes you track about your audience

Access Note

The Molds tab is access-controlled. In current master it is tied to enterprise-manager/support-style access, and related feature flags can also affect availability.

---

Populations

Empty State Still Exists

The legacy /audience/* route now redirects into /people/*.

Populations is a real page in current master, but you may still see an empty state with the message:

"Stay tuned! We're working on unveiling the new populations feature in your workspace."

This is expected when your workspace has no population data yet. It does not mean the entire Populations feature is missing from the product.

When population data exists, the page supports the normal population management flow: list/grid browsing, search and sorting, detail pages, persona views, and analytics.

Current Product State

• Populations has a real route and detail experience in current master
• the empty state can still appear when there are no populations to show
• older docs that describe Populations as broadly unavailable or hidden behind /audience are outdated

Why Use Populations?

Imagine you're launching a new skincare line and want to hear from 200 premium beauty consumers — but your panel only has 50. Populations let you create AI representations of your target audience so you can:

• Test campaign questions before sending them to real participants
• Get diverse perspectives across age, location, and lifestyle
• Validate concepts quickly without waiting for recruitment
• Reduce costs by using AI for initial rounds of feedback

Browsing Your Populations

Populations appear in a card grid. Each card shows the population name, member count, and when it was last updated. Use the search bar to filter by name, or sort by most recently updated.

Click the three-dot menu on any card for the current population actions:

• Show Details
• Take Campaign (requires aiSimulationEnabled feature flag)
• Change visibility (available for enterprise managers)
• Delete (requires admin role)

Population Details

Click any population card to see detailed analytics about that audience group.

The details page includes:

• Persona Carousel — Browse representative personas with their avatar, name, and key demographics
• Charts View — Visual breakdowns of your population by demographics, preferences, and behaviors (donut charts, bar charts, treemaps, and more)
• Table View — A detailed list of all personas with sortable columns

In the attached-persona table, each row also has a contextual menu with:

• Conversation
• View Details
• Remove from Population (where that action is available)

Use the category pills to filter which demographic facets are displayed. Hover over any chart element for exact counts and percentages.

Understanding Population Analytics

The charts view provides rich visual data about your population. Here's how to read each chart type:

Donut Charts

Donut charts show the proportional breakdown of a single attribute (e.g., age range, gender, location). Each colored segment represents one value. Hover over any segment to see the exact count and percentage.

What to look for:

• Dominant segments — large segments may indicate your population skews toward certain demographics
• Balance — for representative research, check that no single segment overwhelms the rest
• Missing segments — small or absent segments may mean your population doesn't cover certain demographics

Bar Charts

Bar charts compare values across categories (e.g., product preferences, interest areas). Bars are typically sorted from highest to lowest.

What to look for:

• Relative sizes — quickly see which attributes are most common in your population
• Distribution shape — a flat distribution means even spread; a steep drop-off means concentration
• Outliers — unusually tall or short bars may indicate interesting population characteristics

Treemaps

Treemaps display hierarchical data as nested rectangles. Larger rectangles represent more common attributes. They're especially useful for visualizing many categories at once.

What to look for:

• Dominant categories — the largest rectangles are the most common values
• Variety — many small rectangles indicate high diversity in that attribute
• Groupings — related values may cluster together visually

Comparing Populations

To compare two populations, open each in separate browser tabs and look at the same chart type side by side. This is useful for understanding how different target audiences differ in demographics or preferences.

Creating Populations

Populations are typically created during campaign setup or through the Agent Builder. The Populations tab is primarily for viewing and managing existing populations.

Best Practice

Keep populations focused on specific demographics or use cases. Use clear names like "US Millennials - Urban - Tech Enthusiasts" so your team can quickly find the right audience.

---

Humans

The Humans tab is where you manage your real participants — the people who receive campaign invitations and provide genuine video feedback.

Viewing Your Contacts

Contacts appear in a sortable table with columns for Name, Age, Last Active, and action options. Click any column header to sort.

Use the search bar to find contacts by name. Click the filter button to narrow results by campaign participation or custom property values.

Build Richer Profiles Over Time

Combine multiple filters to uncover insights — for example, find participants who completed your holiday campaign AND have a "Gold" loyalty tier, or locate consumers in a specific region who haven't been surveyed recently.

Adding Contacts

Click + Add to open the current Add Creators modal. You can paste one or more email addresses, then either add them directly or use the modal's Add to list dropdown to place them into an existing list first.

Bulk Actions

Select multiple contacts using the checkboxes, then use the bulk action buttons to:

Action	What It Does
Delete	Permanently remove selected contacts and their response history
Add to List	Add contacts to an existing list or create a new one
Add Property	Assign a property value to all selected contacts at once

The Humans table also has a row-level menu on each contact for contextual actions such as Delete/Remove and Add Property. Delete actions use a confirmation step.

The Add Property flow opens a modal where you choose the property, choose the value, and decide whether to Add value or Replace value.

Contact Profiles

Click a contact's name to open their full profile.

A profile includes:

• Basic Information — Name, email, avatar, age, gender, location
• Stats Cards — Total responses, Video minutes, and Engagement metrics
• Summary tab — Overview of the contact's activity and attributes
• Interests tab — Topics and categories the contact has shown interest in

Contact Profile: Deep Dive

Each contact profile is organized into sections that give you a 360-degree view of the participant.

Basic Information

The header area shows the contact's name, email address, avatar (if available), age, gender, and location. This information may come from their account registration or from properties you've imported.

Editing Contact Information

Click on any editable field to update it directly. Changes are saved immediately. Only users with Edit permission on the People section can modify contact information.

Custom Properties

All custom properties assigned to this contact appear in a organized list. Each property shows its key and current value. You can edit property values directly from the profile page.

Campaign History

A chronological list of every campaign this contact has participated in, showing:

• Campaign name and date
• Completion status (completed, partial, not started)
• Number of questions answered

Click any campaign entry to jump to that contact's specific responses.

Response Data

For each completed campaign, you can view:

• Video responses — watch the recorded video directly in the profile
• Text transcripts — read the AI-generated transcript of video responses
• Written answers — see responses to text-based questions
• Timestamps — when each response was submitted

Segment Memberships

Shows all lists and segments this contact belongs to. This helps you understand how this person fits into your audience organization and which campaigns they might be targeted for next.

How Contacts Join Your Workspace

Source	How It Works
Email Invitation	Direct invite from the Humans tab
Campaign Links	Public or shared campaign links
CSV Import	Bulk import via Properties upload
QR Codes	Scanned from physical materials

---

Lists & Segments

Organize your contacts into reusable groups for campaign targeting, analysis, and ongoing engagement. Use the dropdown at the top to switch between Lists and Segments.

Lists vs. Segments

	Lists	Segments
Type	Static	Dynamic
How membership works	You manually add and remove contacts	Contacts are automatically included based on rules
Updates	Only when you make changes	Automatically as contact data changes
Best for	VIP panels, project teams, hand-picked groups	Behavioral targeting, demographic filters

Working with Lists

Lists are manually managed groups. Click New List to create one, then add contacts from the Humans tab using bulk actions.

The list table shows Name, User count, and Created date. Click any list name to edit it inline. Use the three-dot menu to edit, duplicate, or delete a list.

Bulk actions when multiple lists are selected:

• Delete — Remove all selected lists
• Combine Lists — Merge selected lists into a single new list

Use Case

Create a list called "2026 Holiday Research Panel" for a seasonal campaign, or maintain a "VIP Beauty Enthusiasts" list of your most engaged participants who get early access to new studies.

Working with Segments

Segments are dynamic groups that automatically update as contacts match your rules. Click New Segment to open the segment-builder modal.

Building a segment:

1. Choose a Property to filter on (age, location, custom attributes, etc.)
2. Select an Operator — Equals or Not Equals
3. Enter the Value to match
4. Choose Any (OR logic — match if any rule is true) or All (AND logic — must match every rule)
5. Preview matching contacts and save

Example segments for a beauty brand:

Segment	Rules	Use Case
Premium Skincare Consumers	purchase_frequency > 4/year, category = "skincare"	High-value product research
Gen Z Beauty Enthusiasts	age between 18-27, interest contains "beauty"	Trend research
Dormant Participants	last_active > 90 days	Re-engagement campaigns
Northeast US Panel	region = "Northeast"	Regional product testing

Each segment row uses a contextual menu with Edit, Copy, and Delete. Bulk actions across selected segments currently include Delete and Create Combined Segment, and both single-delete and bulk-delete flows use confirmation modals.

Segment Builder: Complete Examples

Here are detailed segment rule configurations for common research scenarios. Use these as templates and adapt the property names to match your workspace.

1. High-Value Beauty Consumers

Goal: Identify your most engaged, high-spending beauty consumers for premium product research.

Rule	Property	Operator	Value
Rule 1	purchase_frequency	greater than	6
Rule 2	category	contains	skincare
Rule 3	satisfaction_score	greater than or equal to	4
Logic	All (AND) — contact must match every rule

Result: Contacts who buy skincare products more than 6 times a year AND report high satisfaction.

2. At-Risk Customers

Goal: Find customers who may be disengaging so you can target them with re-engagement campaigns.

Rule	Property	Operator	Value
Rule 1	last_purchase	greater than	90 days ago
Rule 2	satisfaction_score	less than	3
Logic	All (AND)

Result: Contacts who haven't purchased in 90+ days AND have low satisfaction scores. These are prime candidates for win-back research.

3. Gen Z Trendsetters

Goal: Build a panel of young, digitally engaged early adopters for trend and concept testing.

Rule	Property	Operator	Value
Rule 1	age	between	18 and 27
Rule 2	social_media_engagement	equals	high
Rule 3	early_adopter	equals	true
Logic	All (AND)

Result: Young consumers who are highly active on social media and tend to adopt new products early.

4. Regional Research Panel

Goal: Create a geographic panel for region-specific product testing or ad campaign evaluation.

Rule	Property	Operator	Value
Rule 1	region	equals	Northeast
Rule 2	language	equals	English
Rule 3	available_for_research	equals	true
Logic	All (AND)

Result: English-speaking contacts in the Northeast who have opted in for research participation.

5. Lapsed Participants

Goal: Re-engage experienced participants who haven't been surveyed recently.

Rule	Property	Operator	Value
Rule 1	last_survey_completed	greater than	180 days ago
Rule 2	total_surveys	greater than or equal to	3
Logic	All (AND)

Result: Contacts who have completed at least 3 surveys historically but haven't participated in 6+ months. These people know the platform and are likely to provide quality responses if re-engaged.

6. Multi-Category Shoppers

Goal: Find consumers who shop across multiple product categories for cross-selling research.

Rule	Property	Operator	Value
Rule 1	categories_purchased	contains	skincare
Rule 2	categories_purchased	contains	haircare
Rule 3	categories_purchased	contains	makeup
Logic	Any (OR) — match if at least two are true

Segment Rule Logic

• All (AND): Narrows your audience. Every rule must be true. Use when you need a very specific segment.
• Any (OR): Broadens your audience. At least one rule must be true. Use when you want to cast a wider net.
• Combine both: Start with AND for your core criteria, then create additional segments with OR logic for broader targeting.

---

Properties

Properties are custom attributes you can track for contacts beyond the built-in fields like name, email, and age. Define any data point relevant to your research — purchase history, loyalty tier, product preferences, or anything else.

Property Types

Type	Description	Example Use
Text	Free-form text	Occupation, favorite brand
Number	Numeric values	Household income, satisfaction score
Date	Calendar date	Last purchase date, signup date
Single Select	Choose one option from a list	Membership tier (Gold/Silver/Bronze)
Multi-select	Choose multiple options	Product categories of interest
Boolean	True/false toggle	Has purchased, opted into marketing

Creating Properties

Click + New Property to open the property modal, then fill in the name, type, and optionally a category to keep things organized. Property rows also use a three-dot menu for Edit and Delete, and delete actions confirm before they run.

Property Management Best Practices

Standard Properties Every Workspace Should Have

Start with these foundational properties. They support the most common segmentation and targeting use cases:

Property	Type	Category	Why It Matters
loyalty_tier	Single Select (Gold/Silver/Bronze)	Purchase Behavior	Segment by customer value
purchase_frequency	Number	Purchase Behavior	Identify heavy vs. light buyers
last_purchase_date	Date	Purchase Behavior	Find lapsed customers
preferred_category	Multi-select	Preferences	Target by product interest
satisfaction_score	Number (1-5)	Feedback	Track sentiment over time
research_opt_in	Boolean	Research	Know who's available for studies
source	Single Select	Demographics	Track how contacts joined
region	Single Select	Demographics	Geographic targeting

Naming Conventions

• Use snake_case for property names (e.g., purchase_frequency, not Purchase Frequency or purchaseFrequency). This keeps things consistent and prevents issues with CSV imports.
• Keep names short but descriptive — loyalty_tier is better than customer_loyalty_membership_tier_level.
• Use the same property name across all imports. If one CSV uses "region" and another uses "location," you'll end up with two separate properties.

Category Organization

Group your properties into logical categories to keep the Properties page manageable:

Category	What Goes Here
Demographics	Age, gender, region, language, income
Purchase Behavior	Frequency, recency, spend level, loyalty tier
Preferences	Product interests, brand preferences, communication preferences
Research	Opt-in status, last surveyed date, total surveys completed
Feedback	Satisfaction scores, NPS, complaint history
Campaign	Source, acquisition date, campaign participation

When to Use Each Property Type

If you need to...	Use this type
Store a category with one answer (e.g., "Gold" or "Silver")	Single Select
Allow multiple answers (e.g., interested in skincare AND haircare)	Multi-select
Track a number for calculations or comparisons	Number
Record a date for time-based segments	Date
Store open-ended info like notes or occupation	Text
Track a simple yes/no (e.g., opted into marketing)	Boolean

Managing Categories

Organize properties into categories like "Demographics", "Purchase Behavior", or "Research Preferences" using the category dropdown. Click + Create new category or Edit Categories to open the category-management modals.

Inside an individual property-value table, the current UI also exposes row-level Delete actions and a bulk Add to list flow.

Importing Data with CSV

Upload properties and values in bulk:

1. Click the upload button
2. Select your CSV file
3. In the CSV Upload modal, choose Add values or Replace values
4. Upload the file

The modal also includes a Download example CSV file shortcut.

CSV Import: Format Specification

The current Properties CSV upload expects a narrow three-column file, not a wide spreadsheet with one property per column.

Required Columns

Column	Required?	Description
EMAIL_OR_CREATOR_TAG	Yes	The contact identifier used to match the person
ATTR_NAME	Yes	The property name that already exists in Vurvey
ATTR_VALUE	Yes	The value to add or replace for that property

Format Rules

Rule	Details
Encoding	UTF-8 (most spreadsheet apps export this by default)
Header row	First row must contain column headers
One property per row	Each row maps one contact to one property/value pair
Existing property names	ATTR_NAME should match a property that already exists in your workspace
Single file	The upload modal accepts one CSV at a time

Sample CSV

EMAIL_OR_CREATOR_TAG,ATTR_NAME,ATTR_VALUE
jane@example.com,loyalty_tier,Gold
jane@example.com,preferred_category,skincare
john@example.com,loyalty_tier,Silver
john@example.com,research_opt_in,true

Common CSV Errors and Fixes

Error	Cause	Fix
"No matching contacts found"	Email addresses don't match existing contacts	Check for typos, extra spaces, or different email domains
"Unknown property"	ATTR_NAME does not match an existing property	Create the property first, or fix the property name spelling
Garbled characters (?, ÃÂ)	Wrong file encoding	Re-save as UTF-8 in your spreadsheet app
Extra rows imported	Hidden rows or trailing blank rows	Delete empty rows at the bottom of your spreadsheet

CSV Import Best Practices

1. Test with a small file first — try 5–10 rows before importing hundreds.
2. Use "Add values" first if you're appending data to existing contacts.
3. Use "Replace values" carefully because it overwrites the targeted values.
4. Open in a text editor to verify formatting if your spreadsheet app adds unexpected characters.

Deleting Properties

When you delete a property, all values assigned to contacts are also removed. Any segments using that property will need to be updated. Check the "Used" count before deleting.

---

Molds (Enterprise)

Access-Controlled Feature

Molds are not available in every workspace. In current master, access depends on the mold-related access path and feature configuration for your workspace.

Molds are reusable persona templates for consistent AI population generation. Design a mold once — defining demographic parameters, behavioral characteristics, and generation rules — then use it to create unlimited AI personas with consistent attributes.

Mold Lifecycle

Status	Description
Draft	Work in progress, preview only
Live	Published and ready for population generation
Archived	Deprecated but retained for reference

Mold Details

Click any mold to open its configuration page.

From here you can edit the mold's name, description, category, status, and configuration. The system validates readiness before publishing — checking required fields, valid configuration, and demographic coverage thresholds.

---

Integration with Campaigns

Your People data connects directly to campaigns, creating a powerful targeting and analysis loop.

Targeting Segments When Launching Campaigns

When you create or launch a campaign, you can select specific lists or segments as your target audience:

1. Create your campaign in the Campaigns section.
2. During setup, choose your target audience — select from your existing lists or segments.
3. Only contacts in the selected group receive the campaign invitation.
4. As the campaign runs, new contacts who join the segment (in dynamic segments) may also be included.

Targeting Strategy

• Use segments for ongoing campaigns — dynamic membership means new qualifying contacts are automatically included.
• Use lists for one-time studies — static membership ensures a fixed group receives the invitation.
• Combine both — target a segment for broad reach, then create a list of the most engaged respondents for follow-up studies.

Using Populations for AI Testing

Before launching a campaign to real participants, test it with AI populations:

1. Create an AI population that matches your target demographic.
2. Run your campaign against the population to collect simulated responses.
3. Review the simulated results — Are the questions clear? Are responses diverse? Do you see the insights you expected?
4. Refine your campaign based on what you learn, then launch to real contacts.

Tracking Participation Across Campaigns

Use the Humans tab and contact profiles to track how individual contacts participate over time:

• Filter by campaign to see who participated in a specific study.
• Check individual profiles to see a contact's full campaign history.
• Create segments based on participation patterns (e.g., "completed 3+ campaigns" or "last surveyed more than 90 days ago").
• Avoid survey fatigue by checking how recently contacts were last surveyed before targeting them again.

---

Common Workflows

Build a Research Panel for a New Product Launch

1. Import contacts — Upload a CSV with demographic data via Properties
2. Add custom properties — Create attributes like "product_interest", "purchase_frequency", and "loyalty_tier"
3. Create segments — Build dynamic groups like "Premium Skincare Buyers" or "Available for Research" (last surveyed > 30 days ago)
4. Launch campaigns — Target your segments when you're ready to collect feedback

Test Campaign Questions with AI Before Going Live

1. Create an AI population during campaign setup that matches your target audience demographics
2. Run the campaign against the AI population to collect simulated responses
3. Review the results — Are the questions clear? Are you getting diverse perspectives?
4. Refine and relaunch to real participants with confidence

Segment Your Beauty Enthusiasts by Purchase Frequency

1. Create properties for "purchase_frequency" (Number) and "product_category" (Multi-select)
2. Import data via CSV from your CRM
3. Build segments like "Heavy Buyers" (purchase_frequency > 8/year) and "Occasional Shoppers" (purchase_frequency 1-3/year)
4. Compare insights — Run the same campaign to both segments and see how responses differ

---

Troubleshooting

Contacts Not Appearing

• Verify the invitation was sent and the email address is correct
• Ask the contact to check their spam folder
• Make sure you're viewing the correct workspace
• Try refreshing the page

Segment Returns Unexpected Results

• Review each rule carefully — check that Any vs. All logic is correct
• Verify that contacts have the expected property values set
• Test with a known contact to confirm matching behavior

CSV Import Fails

• Save as UTF-8 encoding
• Match column headers exactly to property names
• Use YYYY-MM-DD format for dates
• Try a small test file first (5-10 rows)

Duplicate Contacts

1. Search for the contact by email to confirm the duplicate exists.
2. Duplicates usually happen when the same person is invited via different methods (email invite + campaign link) or when email addresses have slight variations (e.g., jane@company.com vs. jane+research@company.com).
3. To resolve, decide which profile has better data, then manually add any missing properties from the duplicate to the primary profile.
4. Delete the duplicate contact.
5. To prevent future duplicates, standardize on one email address per contact and use CSV imports with the email column to match existing records.

Property Values Not Showing Correctly

1. Check that the property type matches the data. For example, if you created a "Number" property but imported text values, the data may not display correctly.
2. For Single Select properties, verify that the imported value exactly matches one of the defined choices (including capitalization).
3. For Date properties, confirm the format is YYYY-MM-DD.
4. If values appear garbled, the CSV file may not be UTF-8 encoded. Re-export from your spreadsheet with UTF-8 encoding.
5. Try editing the value directly on a contact profile to see if it saves correctly — this can help isolate whether the issue is with the import or the property definition.

Segment Not Updating Dynamically

1. Confirm the segment uses dynamic rules (segments, not lists). Lists are static and only update when you manually change them.
2. Check that the property values for your contacts have actually changed. The segment only updates when the underlying data changes.
3. Verify the rules still make sense — if you changed a property from Text to Number type, existing rules may not match anymore.
4. Try editing the segment, making no changes, and saving again. This can trigger a refresh.
5. If a segment shows 0 contacts when you expect more, review each rule individually to find the one that's too restrictive.

CSV Import Column Mapping Issues

1. Column headers must match property names exactly — including capitalization, spaces, and special characters.
2. Remove any extra spaces before or after header names (common in Excel exports).
3. If you have columns in your CSV that don't match any property, they'll be ignored silently. Check the import summary to see which columns were mapped.
4. For multi-select properties, use semicolons to separate values (e.g., skincare;haircare), not commas (which would be interpreted as a CSV delimiter).
5. If column mapping seems random, open your CSV in a text editor to check that it's actually comma-delimited and not tab-delimited or semicolon-delimited.

Population Charts Showing Unexpected Distributions

1. Check the population size — very small populations (under 20 personas) may not show representative distributions.
2. Verify the population was generated with the correct demographic parameters. If the source data was skewed, the population will reflect that skew.
3. Use the category pills to filter which attributes are displayed — sometimes an unexpected distribution in one chart makes sense when viewed alongside other attributes.
4. If the population doesn't match your target demographic, consider regenerating it with more specific parameters.

---

Frequently Asked Questions

Click to expand

Q: How many contacts can my workspace hold? There's no hard limit on the number of contacts. Workspaces with thousands of contacts work well. If you have very large lists (10,000+), filtering and search become especially important for finding the right people.

Q: Can I export my contact list? Currently, contact export is handled through your workspace settings or by contacting support. Individual contact data can be viewed in profile pages.

Q: What's the difference between a population and a list? A population is a group of AI-generated personas — they don't represent real people. A list is a manually curated group of real human contacts. Use populations for quick AI testing; use lists for targeting real participants.

Q: Can I use the same contact in multiple campaigns? Yes. A contact can participate in as many campaigns as they're invited to. Use the "last surveyed" property and segments to avoid over-surveying.

Q: How do I prevent survey fatigue? Track the last_survey_completed date as a property. Create a segment rule like "last_survey_completed > 14 days ago" to exclude recently surveyed contacts from new campaigns.

Q: Can I merge two lists? Yes. Select multiple lists, then use the Combine Lists bulk action to merge them into a single new list.

Q: Do segments update in real time? Segments update as contact data changes. When you update a property value (manually or via import), any segments using that property will reflect the change.

Q: What happens when I delete a contact? The contact and all their response data are permanently removed. They'll also be removed from all lists and segments. This action cannot be undone.

Q: Can I re-invite a deleted contact? Yes, but they'll start fresh as a new contact. Their previous response history will not be recovered.

Q: How do populations work with campaigns? During campaign setup, you can select an AI population as your target audience. The population's AI personas will "respond" to your campaign questions, giving you simulated data that you can analyze before launching to real participants.

---

Next Steps

• Launch a campaign targeting your audience
• Use AI agents to analyze research data
• Automate audience workflows
• Upload data to datasets for AI analysis