Custom AI models for Predictors and Indicators

Then query it all using ChatterBox, Slack, Teams or API calls.

Use Cases

Ziggy can integrate custom models that are trained on your own data to perform many tasks to make your operations more efficient.

Integrate them with any front end, including Slack, Teams, ChatterBox and your own front end for instant results on an individual record.

Perform database operations to update data and systems so your analytics tools can utlilize results across entire datasets.

Sales & Marketing

Churn predictions
CLV, CAC etc.
Next purchase / reorder probability
Deal close probability
Lead scoring & prioritization
Upsell / cross-sell recommendations

Customer Support

Ticket triage & routing models - department routing, urgency
Proactive support alerts
Resolution time prediction
Sentiment analysis

Operations & Supply Chain

Demand forecasting
Inventory optimization - optimal quantities, reorder points
Predictive maintenance - lined to sensor and maintenance history

Finance

Credit risk modeling
Fraud detection
Forecasting & budgeting

HR & Talent

Candidate screening
Employee attrition prediction
Shift staffing prediction
Training effectiveness modeling

Product R&D

Feature adoption prediction
Usage pattern clustering
Quality defect prediction

Model	Business Goal	Training Data Needed	Expected Impact
Predictive Lead Scoring	Identify high-value leads to prioritize sales outreach	Historical CRM data: lead source, industry, company size, engagement metrics, deal outcomes	+10–30% sales conversion rate, reduced wasted outreach
Churn Prediction	Retain customers before they leave	Subscription data, product usage logs, support tickets, NPS/survey scores	Reduce churn by 10–20%, protect recurring revenue
Demand Forecasting	Optimize inventory & staffing based on future demand	Historical sales, promotions, seasonality, market trends	Reduce stockouts/overstock by 15–30%, improve cash flow
Fraud/Anomaly Detection	Spot suspicious transactions or activities in real time	Transaction logs, historical fraud cases, customer behavior data	Prevent fraud losses, protect brand trust
Predictive Maintenance	Prevent costly equipment failures	IoT/sensor readings, maintenance logs, repair history	Reduce downtime 20–50%, extend asset lifespan

Churn Prediction

This is one of the most common uses cases for a custom trained model. The key is to have enough historical sales data for the model to learn with.

You can bring your own model as well and then hook that into the rest of the Ziggy ecosystem.

Hover for full screen

Watch a user retrieving churn probability and other indicators from a custom model.

A custom trained model is much more reliable than using a generic platform. Your exact needs are catered for and so results will be more accurate. They key thing is to have enough quality data to train on.

Process - a discovery call is required to understand your data and your precise requirements. We then set up Ziggy Flows to feed your data to the model for training.

your model can be constantly kept up to date using Ziggy Flows so you never have to worry about retraining.

Model Training

We provide a fully customized training module. This is then loaded with data from a Ziggy Flow.

Everything about your model is tailored to the available data fields you have and what you need to get out of the model.

We use a Ziggy Flow to pass data for train the model. Another Ziggy Flow can be used to keep the model up-to-date as your data accumulates.

This can be tuned to discard older data so it retains accuracy as your business evolves.

Integration and Front Ends

You can access your model from your CRM/ERP, Slack, Teams, ChatterBox or any other system that can make an API call.

This is ideal for CRM cards or anywhere you want to pass a record of data and retrieve the predictors and other model outputs (yellow box).

Bulk Updating

Use Ziggy's scheduler or trigger a Flow with an API call to update databases, CRM/ERPs etc. When you run your reports, your data will contain up-to-date data from the model. This can be done incrementally or in bulk.

This Flow recalculates all indicators on demand and updates any platform or database. It can restrict updates to underlying changes in your data to keep things fast.

Now, when you run reports or view data, the Churn % is shown. This is great when running any sort of analytics aggregations in any reporting tool.