Syntra's AI Agent works for you - making and answering calls, sending messages and taking bookings on your behalf. This page explains how we build and operate that AI responsibly: the principles behind it, the human oversight you keep, its accuracy and limitations, how we handle data and training, and the safeguards that keep it safe, fair and transparent.
1.Our principles
Syntra is a hyper-intelligent business companion. Our AI Agent makes and answers calls, sends messages, takes bookings and handles enquiries on your behalf, so the way we build and run that AI matters as much as what it can do. These are the commitments that guide every decision we make.
- Helpful - the AI Agent exists to serve you and your End Users. We build it to be genuinely useful, to work within the scope you set, and to hand off gracefully when it cannot help.
- Safe - we design guardrails, prohibited-use controls and escalation paths so the Services behave predictably and avoid foreseeable harm.
- Transparent - we are clear about what the AI Agent is, what it can and cannot do, and where its Output should be reviewed before it is relied upon.
- Under human control - you configure it, you can review it, and you can switch it off. The AI Agent supports human judgement; it does not replace it.
- Fair - we work to reduce harmful bias and to treat the people who interact with the Services consistently and respectfully.
- Private and secure - we protect your data and your End Users' data, and we hold our AI systems to the same security standard as the rest of our platform.
2.How the AI Agent works
The AI Agent combines leading third-party foundation models with your own configuration, knowledge and guardrails. The underlying models provide general language and reasoning ability; you provide the context that makes the AI Agent yours - your scripts, prompts, knowledge base, tone, pricing, availability, and the rules for what it should and should not do.
The AI Agent acts within the scope you set. It draws on the knowledge you give it to answer enquiries, make and answer calls, send messages, and take bookings, and it follows the guardrails you configure. Where a request falls outside that scope, or is uncertain or sensitive, it is designed to say so, decline, or escalate rather than improvise.
Because part of the intelligence comes from third-party model providers, we choose those providers carefully, keep your use of the Services within their terms and safety policies, and remain accountable to you for how the Services behave overall.
3.Human oversight & control
You are always in control of the AI Agent. Nothing goes live until you set it up, and you can change or stop it at any time.
- You configure it - you decide what the AI Agent knows, how it speaks, which channels and hours it operates on, and the boundaries it must respect.
- You review it - transcripts, recordings and summaries let you see what the AI Agent said and did, so you can refine its behaviour over time.
- You can disable it - you can pause or switch off the AI Agent, or individual capabilities, whenever you choose.
- It can escalate to a human - the AI Agent is built to recognise when a matter should be handed to a person, and to route it to your team rather than press on alone.
Sensitive or urgent matters can be routed to your team - for example, with risk-aware escalation, where the AI Agent detects signals that suggest a caller needs human help and passes the interaction to you promptly. You decide how these escalation rules are set, and you remain responsible for staffing and responding to them.
4.Accuracy & limitations
The AI Agent is powerful, but it is not infallible. Output is generated using artificial intelligence and can be wrong, incomplete, out of date or unsuitable for a particular situation. You are responsible for reviewing and verifying Output before relying on it, sending it to an End User or acting on it.
Output is not legal, medical, financial or other professional advice, and it does not replace a qualified human where one is required. The Services should not be used for emergency or life-safety communications, or for high-risk decisions, without appropriate human review and oversight - the AI Agent is not a substitute for emergency services.
These limitations, and your responsibilities around them, are set out in more detail in our Terms of Service.
5.Data & training
We treat your data with care. How we collect, use and protect personal data is described in our Privacy Policy, and where we process personal data on your behalf we do so as your processor under our Data Processing Addendum.
We do not use Customer Data that identifies you or your End Users to train machine-learning models made available to other customers, except in aggregated or de-identified form or with your permission. We do not sell your Customer Data. Where the AI Agent relies on third-party model providers, we work to ensure your Customer Data is not used by those providers to train their general models, in line with the arrangements we hold with them.
6.Fairness & bias
AI systems can reflect biases present in the data they learn from. We work to reduce harmful bias in how the Services behave, to treat the people who interact with the AI Agent consistently and respectfully, and to monitor Output quality so we can catch and correct problematic patterns. We select and evaluate the third-party foundation models we rely on with fairness in mind, and we welcome reports of behaviour that seems biased or unfair so we can investigate.
7.Transparency & disclosure
People deserve to know when they are dealing with an automated system. Where the law requires it, End Users should be told that they are interacting with an AI Agent rather than a human, and the AI Agent is designed to support that disclosure.
Because you control how the AI Agent is configured and deployed, you are responsible for making the disclosures the law requires for your communications - including any notices or consents needed for automated messaging or for recording and monitoring calls. We give you the tools to make those disclosures; you decide how they are presented and ensure they meet the requirements that apply to you.
8.Safety & prohibited uses
We build guardrails into the Services to keep the AI Agent within safe and appropriate boundaries, and we prohibit uses that could cause harm. The Services must not be used to deceive, defraud, harass or endanger people, to generate unlawful or abusive content, or for other prohibited purposes.
The full list of prohibited use cases, and the obligations that apply to your use of the Services, are set out in our Acceptable Use Policy. We may act on violations to protect End Users, our platform and the wider public.
9.Security of AI systems
We hold our AI systems to the same security standard as the rest of the platform. Access controls, encryption, monitoring and the other safeguards described on our Security page apply to the data that flows through the AI Agent, and our processing of personal data on your behalf is governed by our Data Processing Addendum. We choose third-party model providers that meet appropriate security and confidentiality commitments.
10.Your responsibilities
Responsible AI is a shared effort. You know your business, your customers and the laws that apply to you, so you decide how the AI Agent is used and remain responsible for that use. In particular, you are responsible for:
- Appropriate use cases - deploying the AI Agent only for purposes that are lawful, suitable for automation, and within the guardrails you have set;
- Disclosures - telling End Users they are interacting with an automated system where required, and giving any notices your communications call for;
- Consents - obtaining and honouring the consents and permissions needed to contact, record or message your End Users;
- Review and oversight - reviewing Output where it matters, keeping a human in the loop for sensitive or high-risk matters, and refining the AI Agent's configuration over time.
11.Governance & continuous improvement
Responsible AI is not a one-off exercise. We review our safeguards, guardrails and model choices as the Services and the underlying technology evolve, we monitor how the AI Agent performs and investigate issues that are reported to us, and we update our controls, documentation and this page as our practices improve. As standards and regulation for AI develop, we aim to meet or exceed them, and to keep the commitments on this page current.
12.Contact us
If you have questions about how we build and operate our AI, or you want to report a concern about the AI Agent's behaviour, please contact us:
SYNTRA LLC
8 The Green, Dover, Delaware 19901
hello@syntra.ai