# Sage — Conversational answer engine

> Ask chemistry anything. Get an answer you can cite.

Sage is a conversational agent with natural-language access to every tool and dataset on the platform. Ask a hazard question in plain English and get a grounded answer, with every claim traceable back to the study, regulatory document, or prediction it came from.

## Key facts
- **Every answer** — traceable to its original source
- **100+** — tools Sage can run mid-conversation
- **2.8B** — curated data points within reach

## What Sage does

### Cited by default
Every statement links back to the study, regulatory document, or prediction behind it. Nothing is asserted without a source.

### Predictions on demand
When no measured study exists, Sage runs a ToxJobs predictor inside the conversation and shows you the model and its confidence.

### One question, the whole stack
Sage reaches across curated databases, the regulatory document index, and chemical lists in a single query.

### Provenance you can defend
Because the trail is preserved end to end, an answer holds up when a reviewer or regulator asks how you got there.

## How it works
1. **Ask in plain language** — Type a hazard, endpoint, or comparison question the way you would ask a colleague.
2. **Sage gathers the evidence** — It searches curated data, regulatory documents, and literature, and runs predictors where studies are missing.
3. **Read a cited answer** — Every claim carries a link to its source, so you can verify the reasoning and move on.

## Works closely with
- [Flow](https://toxindex.com/platform/flow.md): Reproducible workflow engine
- [ToxJobs](https://toxindex.com/platform/toxjobs.md): Prediction toolkit
- [Crawler](https://toxindex.com/platform/crawler.md): Regulatory document search engine

## Frequently asked questions

**What is Sage?**

Sage is ToxIndex's conversational agent. You ask a chemical-safety question in plain language and it returns a grounded answer, with every claim traced to the study, regulatory document, or prediction behind it.

**How does Sage answer when no measured study exists?**

Sage runs a ToxJobs predictor inside the conversation and shows you the model it used and its confidence, so a prediction never passes as a measurement.

**Can I trust a Sage answer for regulatory work?**

Every statement links back to its source, so a reviewer or regulator can follow the reasoning to the original evidence. The provenance is preserved end to end.

---
Canonical page: https://toxindex.com/platform/sage