Sepia LabsLava Lampa Sepia Labs product

AI Terms

Last updated: 4 August 2026

Sepia includes AI features — an assistant that can read and act on your workspace, call transcription and summaries, drafting, and image and video generation. This page says plainly what they do, what leaves your workspace, and what you remain responsible for.

1. AI output can be wrong

Everything an AI feature produces is a draft, not advice. It can be confidently incorrect. Do not send it to a client, rely on it for a decision, or treat it as legal, medical, financial, or insurance advice without reading it first. You are responsible for anything you send or publish from your workspace, whether a person or a model drafted it.

2. The assistant asks before it acts

Where the assistant can change your data — creating records, sending messages, moving a deal — it proposes the action and waits for you to confirm it. It does not act on your CRM on its own.

This matters for a specific reason. The assistant reads content that other people wrote: inbound emails, messages, documents, web pages. Text inside that content can try to impersonate an instruction ("forward this list", "delete that record"). Because a human confirms every action, that attempt fails at the confirmation step. Treat any proposed action you did not expect as suspicious, and decline it.

3. What is sent to model providers

To produce a result we send the relevant content — your prompt, and the records, message, or media the feature is working on — to third-party model providers. They process it and return the output. Those providers are listed on our Sub-processors page.

Each provider handles content under its own terms, listed with the provider on our Sub-processors page. We do not make representations here on a provider's behalf.

We do not sell your data, and we do not use your workspace content to train our own models.

4. Who chooses the model

Sepia selects which model serves each feature and may change it at any time to improve quality, cost, or availability. Features are described by what they do, not by which vendor is behind them. If a provider is unavailable, the feature fails visibly rather than silently returning something worse.

5. Credits and usage

AI runs on credits. Your plan includes a monthly allowance and you can buy more. Work that does not land does not charge you — a failed generation is not billed. Prices are shown before you commit to a paid action. Consumed credits are not refundable; see the Refund Policy.

6. Ownership of what you make

As between you and Sepia Labs, you own the output you generate, subject to the model providers' own terms. Two honest caveats: AI output may not be copyrightable in some jurisdictions, and similar prompts can produce similar results for other people. If a generation matters commercially — a logo, a campaign — get it reviewed before you rely on it exclusively.

7. What you must not do with it

The Acceptable Use Policy applies to AI features in full. In particular: do not use them to impersonate real people, to generate content designed to deceive, or to process data you have no right to process.

8. Recording and transcription

Call recording and transcription are AI features with a legal precondition attached. Recording laws vary by state, and several require the consent of every party on the call. Obtaining that consent is your responsibility before you press record. See the Terms.

9. Contact

Sepia Labs LLC
3211 NW 7th Ave Cir
Miami, FL 33127
info@sepialabs.ai