AI policy

What Slide Practice uses AI for

Slide Practice uses AI for one thing today: turning a coaching-session audio recording into a structured recap (Wins, Insights, Commitments, Next focus). The pipeline is audio, then speech-to-text, then large language model, then email. No other product surface uses AI inference at this time.

What Slide Practice will never do with your data

We will not use your session audio, transcripts, or recap content to train third-party models. Our contracts with our AI sub-processors require the same. We will not sell or rent your content. We will not feed your content into any classifier that decides whether to suspend, flag, or rate-limit your account.

Consent and disclosure

Every recording must be preceded by a recorded consent step from the client. Coaches are responsible for compliance with bilateral and all-party consent recording laws in their and their client’s jurisdictions. If a client withdraws consent, all stored audio for that coach-client pair is deleted immediately and no future recording is authorized.

Human review and oversight

Coaches are expected to review every AI-generated recap before delivering it to a client. The product makes this explicit: the recap is shown to the coach first, with edit and discard controls, and is not sent to the client unless the coach approves it. Slide Practice engineering does not read recap content unless you explicitly attach it to a support request.

Errors and hallucinations

Large language models can produce plausible but incorrect text. Slide Practice makes no representation of accuracy. Coaches are the authoritative voice; the AI is a draft generator. We log generation latency and structured-output validity per recap so we can detect regressions.

Audio retention

Audio is retained for 30 days and then automatically deleted. You can delete any recording sooner at any time. The written recap stays in the client portal until either party deletes it.

Reporting AI harm

If an AI-generated recap contains material that is unsafe, defamatory, or clearly hallucinated, report it via /report/safety. We investigate and respond within five business days.