Ansora AI Transparency Statement
How our AI receptionist works, when we tell callers it's AI, and the choices you and your callers have
Last updated July 5, 2026 · Ansora, Miami, FL
The short version
Ansora builds and runs an AI phone receptionist for local service businesses. We believe people deserve to know when they're talking to an AI and what happens to their words. This statement explains, in plain English, that callers may be speaking with an AI assistant and how we disclose that; how the technology works at a high level; what it can and can't do; how we handle recordings and transcripts; the human oversight behind every agent; and the choices callers have — including asking not to be recorded, asking for a human, or simply hanging up. This is a trust document, not a legal contract. For the full details, see our Privacy Policy and Trust Center.
01Who we are and why this exists
Ansora is a Miami, Florida company that builds, provisions, and maintains a done-for-you AI receptionist for local service businesses — HVAC, plumbing, electrical, dental, law, salons and med-spas, restaurants, auto, real estate, and more. When someone calls one of our client businesses, our AI answers the phone 24/7, books appointments, texts callers back, and helps with follow-up.
We wrote this statement to get ahead of a simple expectation that we think is completely fair: people should know when they're talking to a machine, and they should understand what that machine does with what they say. Rules like state bot-disclosure laws and the EU AI Act are pushing the whole industry in this direction. We're not waiting to be told — this is how we want to operate anyway.
This document is written for two audiences: the business owners who hire us, and the everyday callers who reach one of our AI receptionists. We've tried to keep it honest and jargon-free.
02You may be speaking with an AI — and here's how we tell you
When you call a business that uses Ansora, you may be speaking with an AI assistant rather than a human employee. We don't try to hide that.
Every Ansora agent can begin the call with a short spoken disclosure that tells the caller two things: that the call may be recorded, and that they're speaking with an AI assistant. This disclosure is configurable per client and per state, and it is turned on by default for two-party-consent states. Florida — where we're based — is a two-party-consent state, and that reality is part of why we built the disclosure tooling the way we did.
The client business decides the exact wording and whether recording is on, but Ansora provides the tooling so that giving a clear, upfront notice is the easy, default path. The business is responsible for making sure all legally required notices and consents are given; we make that turnkey.
- Spoken AI + recording disclosure available on every agent, default-on for two-party-consent states like Florida.
- Configurable per client and per state; recording can be turned off entirely.
- Staying on the line after the disclosure is treated as consent to continue — you can also ask not to be recorded or end the call at any time.
03How the AI works, at a high level
Our receptionist isn't magic and it isn't a person. It's a pipeline of well-understood technologies stitched together to hold a natural phone conversation in real time.
In plain terms: when you speak, your audio is converted to text by a speech-to-text system. That text is read by a large language model that understands what you're asking and decides how to respond — including looking up availability or capturing the details of your request. The model's reply is then converted back into a spoken voice and played to you. All of this happens in a fraction of a second, over and over, so the conversation feels natural.
We build each agent from the client's own public information — their Google Business Profile and website — so it answers in the business's voice with the business's real details.
- Speech-to-text: your spoken words are transcribed into text in real time.
- Language model: a large language model interprets the request and generates a response, and writes call summaries.
- Text-to-speech: the response is spoken back to you in a natural voice.
- Orchestration and telephony connect it all to the phone line so it works on a normal call.
04What the AI receptionist can do
The agent is built to handle the everyday front-desk work that a small business phone line needs, around the clock, without missing calls or sending people to voicemail.
It's designed to be genuinely useful on routine calls and to hand off gracefully when a call needs a human.
- Answer every call 24/7 in the business's voice.
- Answer common questions about hours, services, location, and pricing based on what the business provides.
- Book, reschedule, and capture appointment requests onto the calendar.
- Text callers back and run speed-to-lead follow-up.
- Capture the caller's name, phone number, and what they need, and pass a clear summary to the business.
- Send review requests and handle some back-office follow-up.
05What the AI cannot and should not do
An AI receptionist is a helpful front desk, not a professional advisor. It has real limits, and we'd rather be upfront about them than let anyone be surprised.
The agent does not give professional, legal, medical, financial, or other regulated advice. If you call a law firm, a dental office, or a medical practice, the assistant can take your information and get you to the right person — it will not diagnose you, quote legal outcomes, or make professional judgments on the business's behalf.
Like all AI, it can occasionally misunderstand a request or get a detail wrong. That's exactly why human oversight, escalation, and the ability to correct the record are built in — see the sections below.
- No professional, legal, medical, or financial advice.
- Not a substitute for emergency services — for a real emergency, hang up and call 911 or the appropriate emergency line.
- It only knows what the business has given it; it can't speak to things outside that scope.
- It can make mistakes, which is why a human is always in the loop.
06Human oversight and tuning
There is always a human behind the agent. Ansora doesn't just flip a switch and walk away.
Before an agent goes live, we pre-build a draft from the business's public information, the owner does a test call, and the owner approves it. In the first week we actively tune the agent on real calls — listening to how it performs, fixing wording, and closing gaps. We maintain the system on an ongoing basis after that.
The AI is also designed to escalate to a human when it should: for urgent calls, situations it isn't set up to handle, or any time a caller asks for a person. It captures the details and routes the caller or a message to the business rather than pretending it can do something it can't.
- Owner reviews and approves every agent with a test call before go-live.
- Ansora tunes each agent on real calls during the first week and maintains it after.
- The agent escalates urgent or out-of-scope calls, and any caller who asks for a human, to a real person.
07How recordings and transcripts are used
When calls are recorded, we use the recording, transcript, and AI-generated summary to do the job the business hired us for: to operate the phone line, book the appointment, deliver an accurate message, and improve that specific business's own agent so it gets better over time.
For caller data — the inbound phone number, call audio and recording, transcript, summary, and booking details — the client business is the owner (the controller) of that data and Ansora acts as its service provider (processor). We handle it on the business's behalf, not for our own separate purposes. We do not sell or rent personal information.
On the question of AI training, we phrase this carefully because part of it depends on third parties: Ansora does not use your call content to build a general-purpose public AI model of our own, and where our AI providers support it, we use their business and API offerings and configure them so that client content is not used to train their models. Because these are third-party providers operating under their own terms, we can't give an absolute guarantee about what every provider does at every moment — but this is the posture we set and maintain.
- Recordings, transcripts, and summaries are used to run and improve the client's own agent — not repurposed for unrelated uses.
- The client business owns the caller data; Ansora processes it on the business's behalf.
- We don't sell or rent personal information.
- We configure AI providers, where supported, so client content isn't used to train their models — stated conditionally, because we can't bind third parties absolutely.
08The providers behind the technology
We're not a black box. The AI receptionist is built on a handful of specialized providers, each doing one part of the job under its own terms and data-processing agreements.
We keep this list current in our Privacy Policy and Trust Center so you always know who touches the technology.
- Vapi — real-time AI voice agent orchestration.
- Twilio — phone numbers, telephony, and SMS.
- Deepgram — speech-to-text transcription.
- OpenAI and/or other large-language-model providers — language understanding and call summaries.
- Stripe — payment processing (card data is handled by Stripe, not stored by Ansora).
- Supabase — database and authentication hosting.
- Resend — transactional email.
- Google Workspace — internal team email and productivity.
09Accuracy and correcting the record
AI is powerful but not perfect. An agent can mishear a name, transpose a number, or misjudge what a caller wanted. We treat that as a normal part of operating the system, not a rare failure.
Because summaries and booking details go to a real business that then acts on them, mistakes surface quickly and get fixed. If something the AI captured about you is wrong — a misspelled name, an incorrect phone number, a garbled request — it can be corrected. Callers raise corrections with the business (which owns the data), and Ansora assists as the processor. Business owners can correct or update records directly.
We also use real-call feedback to tune the agent so the same mistake is less likely next time.
- The AI can make errors; humans review and act on its output.
- Callers can ask the business to correct information about them; the business can update records directly.
- Real-call tuning reduces repeated mistakes over time.
10Your choices as a caller
You're never trapped in a conversation with a machine. If speaking with an AI isn't what you want, you have straightforward options, and the agent is built to respect them.
After the opening disclosure, staying on the line means you're okay to continue. If you'd rather not be recorded, or you'd rather talk to a person, just say so — or hang up. Any of those is a completely valid choice.
- Ask not to be recorded.
- Ask to speak with a human — the agent will escalate you or take a message for a person.
- End the call at any time.
- Exercise privacy rights (access, correction, deletion, portability) through the business, with Ansora's help as processor.
11A note for healthcare callers
For healthcare clients such as dental, medical, and med-spa practices, any protected health information is handled under a separate Business Associate Agreement (BAA) between Ansora and the practice, with additional protections like PHI-redaction and short or zero call-recording retention options.
In plain terms: if you're calling a healthcare provider that uses Ansora, we handle your health-related information under the stricter rules that apply to that kind of business.
12Where to learn more
This statement is meant to be a clear, honest overview of how our AI works and the choices you have. It isn't the whole story of how we handle data or secure our systems.
For the complete details — including data retention defaults, security practices, your full privacy rights under laws like the CCPA/CPRA and Virginia/Colorado/Connecticut/Utah-style state privacy laws, and our current subprocessor list — see our Privacy Policy and our Trust Center. On our security roadmap, SOC 2 is in progress; we say that plainly and do not claim a certification we don't yet hold. You can reach us any time at hello@ansora.net.
This statement is written in plain English. It is not legal advice, and it should be reviewed by an attorney before being relied on.
- Privacy Policy — how we collect, use, retain, and protect data.
- Trust Center — security practices, subprocessors, and compliance status (SOC 2 in progress).
- Contact — hello@ansora.net.
This document is written in plain English and isn't legal advice. Have it reviewed by an attorney before relying on it. Questions? Email hello@ansora.net.