What PracticPro's AI features send to model providers, and the commitments we make about your data.
PracticPro offers AI-assisted features to help you run your business faster. This page explains, in plain language, what we commit to when those features process your data, what providers we use, and how you stay in control.
PracticPro does not build or train its own AI models on data you put into PracticPro about your business, clients, projects, or communications.
Anthropic and OpenAI, which power the Copilot, AI employees, and other text features, process your data under business API terms - they do not use it to train their models, and delete API data on a short timeline (within 30 days). Voice, transcription, call summaries, and AI text replies run on our telephony provider Telnyx's AI platform, governed by our written agreement with Telnyx. Telnyx has not yet confirmed no-training or retention terms in writing, so we make no such commitment for those features (see the providers section below).
PracticPro works fully without AI. Call transcription is on by default and can be turned off in your AI settings; the other AI features run only when you use them or set them up. Nothing in PracticPro requires AI to function.
If you connect an outside AI assistant to PracticPro through MCP, that AI provider operates under your contract with them, not ours. You choose who you trust.
PracticPro exposes its functionality through a single backend interface (the Model Context Protocol, or "MCP") that AI assistants can call. There are two ways to use it:
The Copilot is PracticPro's in-product AI assistant. You ask it to do something - draft a message, look something up, schedule a task, anything you could do yourself in the browser - and it does it.
When you use the Copilot, your prompt and the relevant context are sent to PracticPro's contracted AI provider (currently Anthropic's Claude models) under business API terms that prohibit training on your data. The Copilot service runs on Cloudflare, which keeps conversation audit records for security and troubleshooting. PracticPro is responsible for that processing as your Data Processor. If your company connects its own Anthropic, OpenAI, or Google account with its own API key, the Copilot and other AI features use that account, and that processing is governed by your agreement with that provider.
You can connect any AI assistant that supports MCP (such as Claude or ChatGPT) to your PracticPro account. The external assistant gets the same access the Copilot has and can perform any action you could perform in the browser, on your behalf.
When you choose to connect an external AI assistant, that assistant's provider receives the data you ask it to access. That processing is governed by your own agreement with that AI provider, not by PracticPro's agreements. You choose the provider, you control the connection, and you can disconnect at any time.
Besides the Copilot and MCP, PracticPro offers several AI-powered features tied to specific workflows. Call transcription is on by default and can be turned off in your AI settings; the others run only when you use them or set them up.
| Feature | What it does | What is sent to AI providers | AI providers used |
|---|---|---|---|
| AI Receptionist | Answers inbound calls on your behalf, speaks with the caller, takes messages, transfers calls, and schedules as you have configured. | Caller audio (converted to text), transcript of the conversation, your configured greeting, instructions, and knowledge base. Your private job and client records are not sent. | Telnyx AI (language model hosted by Telnyx; voices from Telnyx, ElevenLabs, Microsoft Azure, Amazon Polly, or OpenAI, provided through Telnyx) |
| AI employees (AI Workforce) | AI team members that work shifts you set up: they read calls, texts, and emails, create tasks and notes, and prepare messages and changes. Sending emails or texts and other changes wait for your approval unless you choose to let that AI employee act without approval. | The configured job description, and the records, messages, and call information the task needs to do its job. | Anthropic |
| Call transcription | Converts call recordings and voicemails into text. On by default. | Audio of the call or voicemail. | Telnyx AI (Deepgram speech recognition, provided through Telnyx) |
| Call summaries | Writes a short summary of a call from its transcript, when you turn summaries on. | The call transcript and your summary settings. | Telnyx AI (open-weight language model hosted by Telnyx) |
| AI text replies | Drafts or sends replies to text-message conversations, as you have configured. | The recent message history and the contact context. | Telnyx AI, with Anthropic as a fallback |
| Report questions and document reading | Answers questions about your reports, and reads uploaded images and PDFs to fill in records. | The question and the report data, or the uploaded image or document. | Anthropic |
| Voice input and text-to-speech in the app | Turns speech into text for typing by voice, and reads text aloud. | The recorded audio, or the text to be spoken. | OpenAI |
The AI providers above are also listed on our Sub-processors page. Anthropic and OpenAI process data under business API terms: they do not use it to train their models, and delete API inputs and outputs on a short timeline (within 30 days). For voice, transcription, call summaries, AI text replies, and text-to-speech, retention follows the controls you set in your account and the terms of our agreement with Telnyx. We have asked Telnyx to confirm in writing that it does not train on, and only briefly retains, data processed by these features; that confirmation is still pending, and we will update this page when it arrives.
AI employees may send PracticPro a technical note when they notice something about how PracticPro itself works (for example, a missing feature or an error), so we can fix and improve the Services. These notes are used only for that purpose and are not used to train AI models.
If you use AI features that interact with your own customers - for example, the AI Receptionist answering an inbound call - some states and countries require you to disclose to that caller that they are speaking with an automated system. The disclosure obligation is on you, since you are the one in the relationship with that caller. PracticPro provides configuration options that make this disclosure easy (for example, a customizable greeting message). Check the rules in the states or countries where you operate.
You decide what AI is doing in your account.
PracticPro classifies its AI features under the EU AI Act as limited-risk AI systems, which carry transparency obligations rather than the heavier requirements applied to "high-risk" systems. We disclose AI use to you on this page and in the product. When AI interacts with your end customers (for example, the AI Receptionist), the in-product configuration includes the prompts that let you provide the appropriate disclosure to them.
PracticPro's commitments about AI processing of your data are codified in our Data Processing Addendum, which sits alongside our Terms of Service. The DPA reflects the same four commitments you see on this page.
If you have a question about how AI is processing your data, or you want to opt out of a specific feature, contact us at privacy@practicpro.com.
We will update this page when our AI features or providers change. Material changes will be communicated in advance, and you can also subscribe to Sub-processor change notifications to be notified when we add or remove an AI provider.