All your questions.

Is Fantix a data cleanroom?

Fantix is not a data cleanroom. We aim to replace data clean rooms in use cases where probabilistic, synthetic data is nearly as good (or better) than real-world data: by removing the risk of transmitting real user-level data, Fantix can make clean rooms un-necessary in many use cases.

How does Fantix protect personally identifiable information (PII)?

Fantix only ever collects hashed emails (HEMs). Our models are otherwise trained on the premises of the original data owners and no PII other than HEMs is ever transmitted or stored by Fantix.

What is synthetic data?

Synthetic data is artificially generated information that mimics real-world data characteristics. It serves as a privacy-conscious, scalable, and customizable alternative for UX personalization, targeting, and product development without using actual, potentially sensitive, data.

Why is synthetic data better than real world data?

Four reasons: it's 100% privacy-safe because it does not contain PII of any kind; it's accurate because AI is trained on many real world data sets and learns how to debias; it's current because it can be updated frequently; it's scalable, both easier and faster to deploy and not bound by the size of a specific real world data set.

What is CRM data enrichment?

CRM data enrichment is the process of adding, enhancing or refining data in a Customer Relationship Management (CRM) system with additional information from an external data source.

Why is Yellowcake™ better for consumer data enrichment than just buying real world data?

The primary advantages of enriching a data set with synthetic attributes generated by AI are privacy safety and ease of use: synthetic attributes are PII free, and our enrichment model(s) can be queried in real time without PII and without a data clean room.

Does Fantix offer Large Language Models (LLMs)?

No. Fantix is an AI platform for marketing data science. We train models that support predictive analytics, market research, audience enrichment, etc.

How can I query Yellowcake™ without PII?

You can query the model by sending three to seven attributes for each user you want to enrich. To give you a simple example, many e-commerce brands and retailers query the model with just RFM quintiles: for each user, they send just three values (numbers from 1 to 5) indicating which quartile the user is in for Reach, Frequency, and Monetary value. No PII is needed.