Transparency

How AI Works
in Throughline

We use artificial intelligence to power the coaching features in Throughline. This page explains what the AI does, what content is sent to it, and what that means for you.

What the AI does

The AI analyses the structure and clarity of your communication drafts, provides coaching feedback based on Sophie Scott's frameworks, suggests rewrites and alternative phrasing, and checks the logical flow of your narrative. It does not make decisions on your behalf, verify the factual accuracy of your claims, or provide legal, medical, or financial advice.

Which AI we use

We use Claude, a large language model developed by Anthropic, Inc. We access it via Anthropic's commercial API.

What is sent to the AI

When you use the AI coaching features, the text you type into the story fields is sent to Anthropic's API. We do not send your name, email address, or payment information to Anthropic. Each request is processed independently.

AI training

Under Anthropic's API terms, content submitted through the API is not used to train their AI models by default. We have requested a Zero Data Retention agreement and a Data Processing Agreement with Anthropic to formalise this protection. We will update this page when those agreements are in place.

Accuracy

AI outputs can be compelling and wrong at the same time. Before you use any AI-generated content from Throughline, read it, check any factual claims, and satisfy yourself it accurately reflects your intent. You are responsible for everything you publish.

Sophie Scott and Phil Carey's role

Sophie Scott and Phil Carey's communication frameworks are built into the AI's instructions. The AI coaching reflects their approach. The responses you receive are generated by a language model, not by Sophie Scott or Phil Carey personally. They do not constitute their specific advice or endorsement of your communication.

Questions

Contact phil@cornerstonemedia.com.au with any questions about how AI is used in Throughline.