Questions

Predictabl, in plain terms.

What it does, what it needs, and how it differs from the tools you already run.

What is Predictabl?

Predictabl is an agentic data scientist for business data. You describe the outcome you want to predict in plain language, and Predictabl reads your tables, frames the prediction, joins the data, composes features, trains a model and backtests it. The result ships as a live prediction you can call from a product, pipeline, dashboard or AI agent.

What kind of predictions can it make?

Two families. Forecasting future values such as demand, volume or capacity, and classifying future events such as a churn, a late delivery, a no-show or a chargeback. Common cases are operational planning, predictive maintenance, delivery delays and returns, customer churn, refund and chargeback risk, and conversion or retention likelihood.

What data does Predictabl need?

Your existing business tables. It works across multiple related tables and discovers the predictive interactions between them, so there is no need to pre-join the data or hand it a prepared training set. It operates on structured tabular and relational data rather than free text.

How is Predictabl different from a BI tool?

BI describes what happened and diagnoses why, through dashboards, KPIs and historical reporting. Predictabl adds the predictive layer on the same data — forecasts, probabilities and risk scores — and routes them into the workflows and agents where someone can act. It complements a BI stack rather than replacing it.

How is it different from AutoML?

AutoML searches model space over a table someone has already prepared. Predictabl starts from the business question and the raw related tables: it frames the prediction and builds the training set before any model search. Every model is then measured against the honest baseline, such as seasonal-naive or last-value, never against zero.

How long does it take to get a working model?

A validated model in a day rather than a data-science project, which is roughly 60 times faster than the usual path. There is no manual feature engineering step and no ML team required.

How accurate is it?

Models run around 8.5 times more accurate than the baseline a team would otherwise use, and the comparison is always against that honest baseline rather than against zero. On RelArena, Prior Labs' benchmark for prediction on relational data, Predictabl ranked 1st on 4 of 21 tasks and top 5 on 11 of 21 tasks, with no manual feature engineering and no task-specific tuning.

Can I use a Predictabl model inside ChatGPT or Claude?

Yes. Every model is exposed as one versioned endpoint over REST, batch and native MCP, so an MCP-capable assistant — ChatGPT, Claude, Copilot or your own agent — calls the live model and answers from its output instead of guessing. Predictabl is the predictive capability those assistants call, not another chatbot.

Is Predictabl spelled with an e?

No. It is Predictabl, without a trailing e, at predictabl.tech. It is not the English word predictable.

How do I get started?

Book a live walkthrough and bring a question about your own data. The fastest start is a single outcome you already care about and the tables that describe it.

See what Predictabl predicts
on your own data.

Book a live demo