De-identified niche data for training frontier models

Data8 Datasets

We source data from niche markets the open web barely covers, de-identify it end to end, and deliver it to the teams training large models

Dating

Multi-turn, de-identified conversations that help models build rapport, read tone and stay natural across long dialogues

Dating dataset

Relationships

Coaching and counselling transcripts that capture empathy, conflict resolution and long-form dialogue, fully de-identified

Relationships dataset

Events

Planning, ticketing and attendance records that ground agents in real-world scheduling, logistics and group behaviour

Events dataset

The human context frontier models can't find on the open web

Sell us your company

We acquire businesses in niche markets like dating, relationships and events, and turn their data into de-identified training sets for model providers

Sell your company

A fast, confidential conversation with our deals team. Everything we deliver to labs is de-identified, so your users stay protected.

Founders

What we look for

  1. Dating apps and communities
  2. Relationship and coaching services
  3. Event and ticketing businesses
Start a confidential conversation

Refer a company

Know a business that would fit? Introduce us, and you earn a referral fee if the deal closes.

How referrals work

  1. Introduce us to the founder
  2. We lead the conversation
  3. You are paid when the deal closes
Submit a referral
Talk to our deals team