Solutions · Technology

Ship models your team can explain on call.

For product and platform teams running churn, fraud, pricing and growth models. Explain a prediction from a support ticket, compare a retrain with the model it replaces, and get an alert that names the input that moved.

Data
7,043subscribers of a telecom operator, a public sample dataset
Accuracy
0.850 AUCon 1,761 held-out subscribers
One change
78% → 18%the model's answer for Subscriber A on a two-year contract
A new cohort
PSI 5.46when a fibre promotion lands; average P(churn) of the batch is 75%

Where it fits

One explainability layer, from notebook to production.

Growth

Churn, conversion and lifetime value

Why this account is at risk, in terms a retention team can act on, and which change the model responds to most.

Risk

Fraud and abuse scores

Explain a held payment or a blocked signup from the model's structured inputs, and store the explanation with the case.

Fintech

Credit, limits and pay-later decisions

The same reasons and records a bank needs, through an API your product team can call.

Operations

Demand, capacity and arrival-time forecasts

Numeric predictions explained in their own units, and watched for drift.

Platform

One layer across every model

scikit-learn, XGBoost, LightGBM, CatBoost, Keras and PyTorch uploads, or any model behind an HTTP endpoint.

Release

Versions and shadow models

Ship a retrain as a new version, score traffic in shadow, and compare what the two have learned before switching.

A worked example

A churn model, opened up.

Public dataTelco Customer Churn, IBM sample data, via OpenML (IBM).churn-risk-gbmscikit-learn · gradient boosting

A gradient-boosted model predicts whether a subscriber will cancel, from ten account inputs. It is trained on 5,282 subscribers and uploaded with one call; every figure below is what the Kaimei API returned.

churn-risk-gbm · Explain
scikit-learn · gradient boosting
raiseslowers
Telco Customer Churn POST https://api.kaimei.ai/v1/explanations

From a score to an action

What would change the model's mind.

Subscriber A is four months in, on fibre, paying month to month: 78.4% against an average of 25.9%. Kaimei searches for the smallest change that flips the decision, with tenure, dependents and the internet product held fixed.

It finds one: a two-year contract takes the model's answer to 18.5%. For Subscriber C the same change gives 20.0%.

That is a property of the model, not a promise about the customer. It tells the retention team which lever the model responds to, and an experiment tells them whether customers do.

churn.py
from kaimei import Kaimei

client = Kaimei()  # KAIMEI_API_KEY from the environment

client.models.upload(
    model,                      # the fitted GradientBoostingClassifier
    name="Subscriber churn", slug="churn-risk-gbm",
    features=FEATURES, outputs=OUTPUTS,
    X=X_train, y=y_train,       # reference rows
)

explanation = client.explanations.create(
    model="churn-risk-gbm", input=subscriber, method="shap",
    options={"seed": 0, "counterfactual": True,
             "frozen_features": ["tenure_months", "dependents", "internet"]},
)
print(explanation.output["value"])   # 0.7841
client.analyses.create(model="churn-risk-gbm", type="full")

What it learned

The whole model, in a few sentences.

tenure_months is the most influential input, followed by contract and internet. Together they carry 69% of the measured importance (permutation importance over 2000 rows).

2.3% of decisions flip under 5%-σ input noise. Most fragile inputs: monthly_charges, tenure_months.

  • Higher tenure_months decreases the probability of churn, most sharply between 1 months and 5 months (range 12% → 46%).
  • contract = month-to-month pushes the probability of churn highest and contract = two year lowest (range 8% → 30%).
  • monthly_charges has a non-monotonic effect on the probability of churn: it peaks around 103.5 USD (range 23% → 35%).
Global importance · churn-risk-gbmpermutation · 2,000 rows
Tenure28.3%
Contract21.3%
Internet19.9%
Monthly charge10.4%
Paperless billing4.3%
Streaming TV4.1%
Payment method4.0%
Online security3.2%
tenure_months is the most influential input, followed by contract and internet. Together they carry 69% of the measured importance (permutation importance over 2000 rows).

On call

An alert that names what moved.

Five hundred held-out subscribers are logged as normal traffic, and every input is stable. Then a fibre promotion brings a wave of new sign-ups: 500 subscribers with six months or less on fibre, drawn from the same data. The batch averages a P(churn) of 75%, against 26% before. The alert does not say that the dashboard looks odd. It says tenure, internet product, monthly charge and contract have moved, in that order.

Normal trafficstable
Monthly charge0.04
Tenure0.01
Paperless billing0.01
Streaming TV0.01
Payment method0.01
Internet0.01
Recent inputs match the reference population (max PSI 0.04). 500 logged predictions against 5,000 reference rows.
After the promotionsignificant
Tenure5.46
Internet4.24
Monthly charge3.01
Contract2.36
Payment method0.67
Online security0.33
Recent traffic has significant drift (PSI 5.46). Shifted inputs: tenure_months, internet, monthly_charges, contract. 500 logged predictions against 5,000 reference rows.

Population stability index per input. Bars are cut off at 0.5; the line marks 0.25, where Kaimei calls drift significant.

Who it is for

The people who build it, and the people who get asked about it.

01

Machine-learning engineers

A Python client that reads the fitted estimator in your process, plain REST underneath, typed errors and retries. The same four calls for every framework.

02

Product and growth

Open a prediction in the dashboard, change an input and watch the attributions move. No notebook needed.

03

Trust, risk and support

Answer why for a specific account from a stored explanation, by the id your systems already use.

Scope

What Kaimei covers today, and what is coming.

We would rather say so here than have you find out in a trial.

Kind of modelSupported
Yes/no, multi-class and numeric predictions on structured dataYes, uploaded or through your endpoint
Image classificationYes: Grad-CAM for uploaded CNNs; occlusion and LIME for any
Tree models with categorical splits, non-linear SVMs, nearest neighbours, models in other languagesThrough a scoring endpoint, or from attributions you compute. Running them directly, and ONNX uploads, are coming soon
Language models, free text, generationComing soon. Until then, a text model's decision can be explained from structured features you derive
Recommendation and search ranking, detection, segmentation, audio, videoComing soon. Image classification works today
Raw sequences and time seriesComing soon. Until then, a forecasting model on lags and rolling features works

See it on your own model.

Start on the free plan with a model you already have. No sales call required.