How Eyadera uses AI
Eyadera LLC · public disclosure · Version 1.7 (26 Oct 2026) · published at /ai
We use AI models to help us draft, calculate and check our work. People at Eyadera make the judgments and sign off everything you receive. This page says plainly where AI is used and what we promise.
Where we use AI
| Where | What AI does | What a person does |
|---|---|---|
| Eyadera Operations Diagnostic (AI-powered; researches your business from public sources) | Asks the interview questions in an online chat that your owner and one project manager or estimator (both required) answer in their own time; asks short follow-up questions under rules we set; helps organize your exports, find patterns, cross-check interview answers against files, and draft sections of the report | Writes the interview questions and rules, reads every interview, decides what matters, checks every finding against the evidence, edits and approves the report, and runs the sign-off meeting |
| Build work (only if agreed in the order) | Where it fits, AI may be part of what we build for you (for example, drafting a work order for a person to approve), or help us write and test code | Designs, tests and is accountable for the result; AI that drafts business documents is set up so a person approves before anything is sent or posted |
| Our emails to prospects | May help draft outreach text and research public company information | A person at Eyadera approves every email template and checks the first personalised emails before sending starts; after that, emails go out automatically from the approved templates. A person approves every reply we send. Emails go out under "Ethan", a sending name Eyadera uses. Emails are never sent to anyone who has opted out |
What we promise
- AI-powered, reviewed by a person. The Diagnostic is AI-powered: AI prepares the questions, researches your business and asks follow-up questions. We run it only on accounts Eyadera LLC owns, with providers whose terms say they don't train on client data, and a person reviews every page before you see it. AI use is a condition of signing the order form; there is no non-AI version.
- No training on your data. We never use your data to train, fine-tune or improve an AI model, ours or anyone else's, and we use AI providers only through business API services whose terms say they do not train on the data sent.
- Limited retention by providers, stated honestly. Unless zero data retention is in place, providers typically keep inputs and outputs for a limited period set by their terms; each provider's position is on our Approved AI Provider List. For Client data, we use providers under business API terms with no training on client data. Zero data retention has been requested and is being enabled with each provider; no Client data is processed with a provider until it is in place.
- Human sign-off. Nothing an AI model writes in your report or any other Deliverable reaches you without a person at Eyadera reviewing it. The AI interviewer's short follow-up questions appear live, under rules we set, and our outreach emails follow the process in item 8.
- No sensitive data. We ask you not to send health data, ID numbers, payment-card or bank numbers, passwords, drawings, CAD files or export-controlled technical data.
- Accuracy. AI can be wrong. A Diagnostic report is designed so every number is traced to your answers or files, or labelled as an estimate with its basis.
- Interviews you can see. The Diagnostic interview is a written chat, not a call. Before it starts, each person sees a screen saying they are answering an AI interviewer, and nothing starts until they agree. They can stop and come back later. There is no audio or video recording.
- Straight answers about who you are talking to. If you ask whether you are dealing with a person or an AI, we answer honestly. Our outreach emails are drafted with help from software from templates a person has approved; a person checks the first personalised emails before sending starts, and after that they go out automatically from those templates. Every reply we send you is approved by a person. They go out under the sending name "Ethan"; if you ask who Ethan is, we tell you.
- People make the decisions. Approvals, negotiation, signing and payment always stay with a person. Nothing we build or run will approve, negotiate, sign or pay on your behalf; it prepares the work and a person you choose approves it.
- We learn, without keeping your details. Like any advisor, we get better from each piece of work. We keep only general lessons, with your company, your people and your figures removed, and we use them to improve our methods, prompts and templates. They never contain personal data. We store them separately from your data and may also use them to develop our own tools and models, including to train them in future, on accounts we own; we never give them to anyone else to train on.
Which providers
We use AI service providers engaged under business/API terms that prohibit training on Client data, listed on Eyadera's Approved AI Provider List, which is part of the subprocessor list in our Data Processing Addendum (available on request, updated with notice). We may use several providers to review the same work independently. Ask info@eyadera.com for the current list. Any new provider is announced to clients under our Data Processing Addendum (/legal/dpa) before it is used for client data.
Questions
info@eyadera.com
Version 1.7 (26 Oct 2026)