Kinlu turns business questions and customer data into analysis, evaluated models and usable predictions. Without writing the training pipeline yourself.
Start with the question you’re trying to answer, or with the data and model you want to work with.
Kinlu · Conversational experience
Have a business question? Start a conversation.
Describe your objective and provide your data. Kinlu works through data analysis, model building and evaluation, then brings the findings into a business report.
A workflow illustration, not a live run or promised result.
KINLU PredictIllustrative interface
Model Library
Customer segmentation
Clustering · evaluated model
Inspect
Evaluation & cluster profiles
Use
Customer inputs → assigned cluster
Export
Model · Python pipeline · report
Individual predictionCustomer → segment
Uses the saved model and its required inputs.
Your fields and results depend on the trained model.
Kinlu Predict · Model workspace
Know what you want to model? Take the controls.
Upload a CSV or sync Shopify orders. Review eligible use cases, train a model, inspect its evaluation and run individual predictions from the saved result.
Two experiences under Kinlu. Access and data setup are handled within each product; a shared dataset or synchronized session is not assumed.
From a question to a usable result.
The model isn’t the finish line. You need to see what it learned, how it performed and what you can do with it.
Begin with the evidence you have.
In conversational Kinlu, describe your objective and supply a dataset. In Predict, upload a CSV or sync Shopify, then review what the columns can support.
Different questions need different data. A connection alone doesn’t make every model eligible.
Check the model, not just the answer.
Kinlu’s mission workflow builds and evaluates a model for the objective. Predict lets you select eligible models and inspect the completed training results.
Review the evaluation and limitations before relying on any output.
Take something useful into the next decision.
Read the report in conversational Kinlu. In Predict, inspect the model report, download available artifacts and enter the required inputs for an individual prediction.
The report’s detail and available outputs depend on the task and successful training.
Illustrative segmentation journey
Not just an answer. Something you can inspect.
Start with customer data
CustomerPurchase behaviour
Customer AFrequent purchases
Customer BOccasional purchases
Customer CRecent first purchase
Example inputs, not connected customer data.
Inspect the learned groups
Group ACustomer profile
Group BCustomer profile
Group CCustomer profile
Illustrative groups. Actual profiles depend on the model.
Data understandingKnow which inputs support the question.
Model evaluationReview performance and limitations.
A report or predictionUse the output to inform your decision.
Questions worth bringing to your data.
Start with a supported question. Availability depends on the columns, history, targets and quality of the data you provide.
Understand customer groups
Find groups with similar characteristics and inspect their profiles before planning different customer approaches.
Customer segmentation · sufficient useful customer data
Evaluate a churn model
With suitable labelled outcomes, examine how well a model distinguishes churned and retained customers.
Classification · appropriate target and both outcome classes
Explore a revenue forecast
With dated revenue history, evaluate a forecast that can inform planning, not guarantee future sales.
Time series · dates, revenue and sufficient history
Know what you’re starting with.
Practical answers before you bring in your data.
Which experience should I choose?+
Start with conversational Kinlu when you want to describe a business objective and work through an analysis mission. Choose Kinlu Predict when you want direct controls for data readiness, training, model reports and individual predictions.
Do I need to write code?+
You don’t need to write the training pipeline for these workflows. You do need relevant, usable data and a clear question. Some models require a target column or confirmation of how an outcome is defined.
Does connecting Shopify unlock every model?+
No. Shopify orders can supply customer-level data in Predict. Each model has its own requirements. Churn needs suitable outcomes or a valid definition; revenue forecasting needs dated revenue history. Too little data may prevent training.
How autonomous is Kinlu Predict?+
Predict includes a scheduled engine for configured source refreshes, saved-model scoring, conditional retraining and anomaly checks. It runs twice daily, not in real time. These workflows require configuration and enough history; monitoring is still being validated end to end in early access. Your team reviews the outputs and makes decisions.
Are the two workspaces synchronized?+
Don’t assume they are. Each experience has its own access and data setup. This website helps you choose a starting point; it does not transfer datasets or models between products.
Can I take the model out of Predict?+
Successful training can provide a downloadable model artifact, Python pipeline and report. Review the evaluation before using them outside the workspace. Available detail depends on the task and successful artifact generation.