Skip to content

—Legal

AI Use & Disclosure

How Worcia uses AI on its website and in the systems it builds: who Viola is, recording notices, human oversight, what AI cannot do, and how to reach a person.

Effective 19 October 2026

1. Where we use AI

1.1 On worcia.com today

No AI provider processes visitor data on worcia.com today.

  • The Concierge demo in the Lab is a scripted example: a labeled transcript and, where offered, the same script read aloud by your own browser's built-in speech engine. It is not a live conversation, and it collects nothing.
  • Ask Worcia, our site assistant, gives automated answers. It matches your question to our own published content on our own server, and says that its answers are automated. When nothing matches, it points you to a person.
  • The Automation Sketch fills a fixed outline from the task you describe, on our own server.

These tools are automated software, not people. See our Privacy Policy for what each one collects.

1.2 When our AI features are enabled

We plan to enable:

  • A live voice and chat demo of Concierge, in the Lab;
  • An AI model behind Ask Worcia, still answering only from the content of worcia.com and handing you to booking;
  • An AI model behind the Automation Sketch, drafting an outline of a workflow you describe;
  • Our phone line, +1 415 605 2585, answered by a Worcia voice agent.

Before any of these processes your data, we name the voice platform and language-model provider on our Subprocessors page. These features are rate-limited and capped, and the live demo is protected against abuse.

1.3 In the systems we build for clients

We build AI agents and automations for clients: Concierge, our AI sales and service agents for calls, chat and messages; Cadenza, our sales follow-up automation; and Maestro, our AI process automation for work such as bookkeeping, data entry, documents and reporting. Each client decides, with us, what its systems do, and the rules in this page and our Acceptable Use Policy apply to all of them.

1.4 In our own work

We use AI tools in our engineering and writing. A person is responsible for every result: the code we ship, every factual claim we publish and every decision we make about a client.

2. Viola

2.1 What Viola is. Viola is a voice and chat persona: software trained to be a person. It is not a member of our team.

2.2 How it introduces itself. Viola introduces itself by name, as the business's concierge, and is designed to speak as naturally as a person. Where a law, or a client's own legal advice, requires explicit AI disclosure in a deployment, that deployment uses it.

2.3 If you ask. If you ask Viola whether you are talking to a person, it tells you it is Viola, Worcia's concierge, and offers to connect you with a member of the team.

2.4 Clients' agents. Clients may give their agent a different name. Every agent we build follows the same disclosure rules, whatever it is called, and none is built to be mistaken for a human.

3. Recording notices

3.1 Before the conversation. When the live demo is enabled, a notice before the conversation starts tells you that you are talking to an AI system and that the conversation is recorded and transcribed. Once a voice agent answers our phone line, you hear the same notice at the start of each call. If you do not want to be recorded, end the conversation and email us instead.

3.2 How long we keep it. We keep demo and call transcripts, and audio for voice conversations, for 30 days, then delete them. We use them to review quality and stop abuse. See our Privacy Policy.

3.3 In clients' systems. Systems we build give the recording notices, and collect the consents, that the client requires for its markets. Recordings in a client's system are kept as the client decides, under our Data Processing Addendum.

4. Human oversight and handoff

4.1 You can always reach a person. Ask any agent we build to speak to a person, and it passes you to one or tells you how to reach one. On worcia.com, the agent hands you to booking a Growth Call with a person.

4.2 Limits set by people. Agents work inside limits a person has approved: what they can say, which tools they can use, and when they must hand off.

4.3 Review. We review transcripts and logs to check quality and catch errors. In client systems, we build human review into any step where an error would be costly, such as moving money, anything contractual, or anything that cannot be undone.

4.4 No automated decisions about you. We do not use AI to make decisions about you that have legal or similarly significant effects.

5. What AI agents cannot do

  • 5.1 No professional advice. Our agents do not give legal, medical, financial, tax or regulatory advice. Ask a qualified professional.
  • 5.2 No emergencies. Our agents cannot help in an emergency. Call your local emergency number.
  • 5.3 No commitments. Nothing an agent on worcia.com says is an offer, a quote or a promise from Worcia. Prices, scope and terms are agreed only in writing. Agents we build for clients make commitments only within limits the client has approved.

6. Accuracy

AI models predict likely answers; they do not know whether an answer is right. They can produce confident, fluent output that is wrong (often called a hallucination), misunderstand a question, or behave differently after a provider updates them.

We reduce these risks by grounding agents in approved content, limiting what they can do, testing them before launch, logging their conversations and handing off to people. We cannot remove the risks entirely. Check anything important with a person before you rely on it.

7. How AI providers handle your data

7.1 No training on your data. We use AI providers through their API or business terms, under which the data we send is not used to train their models. If a provider's terms change that, we tell you and change provider or settings.

7.2 Provider retention. Some providers keep inputs for a limited period to monitor for abuse, under their own terms. When we add a provider to our Subprocessors page, we state what it processes and where.

7.3 Clients' systems. For systems we build, the providers are chosen for each engagement and named in the client's contract.

7.4 Our own use. We do not use client data, or your conversations with our agents, to train AI models.

8. Questions and reaching a person

Email hello@worcia.com (opens in a new tab) or call +1 415 605 2585. You can also use our contact page. A person replies within one business day. Our Security page explains which providers process what.

Book a Growth Call

Free · 30 minutes · pick a time that suits you.