Your data, and our AI models

No training

Your calls and messages are not training data

Finn does not use Customer Data — call recordings, transcripts, chat messages, prompts, or uploaded contacts — to train or improve any model, ours or a third party’s. Your data is processed only to deliver the service you configured, and for nothing else.

Every model provider in our pipeline is engaged under terms, or with account settings, that exclude your data from model training. This commitment is in our Data Processing Addendum.

Zero data retention

Model providers keep nothing to train on

On Enterprise, our model APIs are configured for zero data retention where the provider offers it: the provider receives a request, returns a response, and does not store the prompt or the output. Otherwise, a provider may keep data for a limited operational period under its own terms (for example, abuse monitoring), and never uses it for training.

What the models receive

Transcript in, response out

Language models usually receive transcript text and prompts, not call audio. The exception is speech-to-speech models, which take call audio directly; they are used only where an agent is configured for them. Speech-to-text and voice providers process only the audio or text needed for their step and return the result.

For chat AI agents, the model receives the conversation so far and the agent’s instructions and knowledge base, and returns a reply. For the Rep Parallel Dialer, speech-to-text listens to the first ~15 seconds of each answered call to detect voicemail, and to recorded calls for live transcripts; the language model receives that transcript text for voicemail detection, summaries, suggested outcomes and coaching, and — for lead briefs — the lead’s first name and public company facts.

The full path — which provider receives which data category — is documented in our data-flow map, and the providers themselves are listed on our subprocessors page.

Retention & deletion

How long data is kept, and when it goes