Eliminate the pain and waste of data science.

Eliminate the cost of asking questions.

Learn more
Watch demo

A new way to do data science


The Grind (CRISP-DM)

Question-first workflow

Today's standard data science workflow (CRISP-DM) is a question-first process designed for today's question-oriented, black-box AI/ML technology. Users first identify a question then set about finding data and technology to answer it. Note that:

  • There is a long distance between question and answer.
  • There are many paths away from answers.
  • The question you answer may not be the one you started with.
  • The only way to know whether, or how well, you can perform is to try things until you are satisfied or give up.

Know thy enemy


The Grind (CRISP-DM)


Zero-Cost Questions

Data-first workflow

Redpoll technology enables a new data-first workflow that is Humanistic. The user provides the machine with data, the machine learns about the data, then the user may ask any number of questions with zero additional effort. Note that:

  • All paths move toward answers.
  • Users can know which questions are answerable before they ask them.
  • Feedback is instant.

See it in action


Zero-Cost Questions

Redpoll AI Platform Features


Increase trust. Reduce iteration and communication overhead.

In high-risk high-impact tasks, people make decisions, not machines. AI must engage with decision-makers directly. reformer is based on cognition research. It learns and stores its knowledge like people do making it fundamentally human-compatible. No black boxes. No surrogates. No translators. Understand what the machine believes and why.

Learn More


Identify and react to anomalies before they do harm.

Detect, analyze, and react to statistical changes in the world before they manifest as failures in production. In addition to detecting anomalous inputs such as data entry errors and attacks, reformer monitors anomalies in its knowledge, helping to maintain safe, reliable decision support.


Continuous insights. Add, edit, and backfill data without retraining.

People do not retrain and neither should AI. reformer learns from streams of data. Users may add new records, add new features and fields, and edit existing data with no retraining or revalidation, minimizing resource use, reducing latency, and ensuring knowledge is always up-to-date.

Learn more

See Core in action

The above demo represents an early GUI prototype run on local hardware. The look, feel, and functionality of the final product will differ.