Start here

ABTO connects AI calls to real user outcomes, showing with data which model earns its cost per feature. Evaluating ABTO? Start with the 'What ABTO solves' tab. Already decided? Go straight to the 'Integration guide' tab.

What ABTO solves Get started

If one of these sounds familiar

Where is our AI cost going?

Break the single invoice line into per-feature cost.

Follow the guideWhy did cost suddenly jump?

Find the day the spike started and the setting that changed that day.

Follow the guideCan we switch to a cheaper model?

Measure the saving first, then check the outcome on 10% of traffic.

Follow the guideCan we switch to the new model?

Screen, validate at 10%, judge, switch. One full loop.

Follow the guideDid the prompt fix actually help?

Run two prompts side by side and let user behavior decide.

Follow the guideIs the AI feature paying for itself?

Report the feature’s worth as one revenue-per-AI-dollar figure.

Follow the guide

A guide for every dashboard screen

Insights

Ask the data questions in natural language

Overview

Manage AI cost and call volume on one screen

Success metrics

Define what success means per feature

Testing

Send trial calls to model and prompt candidates

Routing

Adjust traffic split per variant

Compare

Read variant outcomes side by side

Requests

Drill down to individual calls to find the cause

Settings

Manage integration setup and API keys

Three kinds of data, joined into one flow

User actionask, copy, purchase

Model executionmodel, cost, speed

Product outcomesuccess metric, experiment, decision

The three are joined automatically, down to which response led to a purchase.

Where to start reading

Understand the product

Learn what ABTO does, then the concepts you need to read the dashboard.

Integration guide

The actual integration order: project setup, first request, SDKs, and event design.