Capability without care is not progress.
This page describes our position on responsible technology — how we evaluate the systems we build, where we decline work, and how we treat safety, privacy and human oversight as engineering constraints rather than marketing statements.
Our position
Every technology we build is evaluated against the same three questions: who is affected, what could go wrong, and what would we need to see to be confident it is working as intended. These are not aspirational values. They are conditions we hold ourselves to before, during and after shipping.
We are a small company, and we have no interest in pretending otherwise. Our ethical commitments are calibrated to what a company of our size can actually hold: a narrow set of promises we can keep, rather than a broad set of principles we cannot.
Where we draw lines
Some categories of work we decline, regardless of who asks or what they pay. We do not build systems whose primary purpose is:
- Manipulation, coercion or the removal of informed consent.
- Surveillance of individuals or groups without their knowledge and consent.
- The automation of decisions that fundamentally affect a person's life without meaningful human oversight.
- The presentation of fabricated information as if it were fact.
- Any capability that could only be responsibly deployed by an organisation that has capabilities we do not.
If a project is a poor fit for these standards, we will say so before work begins, not after.
Safety as an engineering constraint
Safety is not a review that happens after a system is built. It is part of the design. This means:
- Failure modes are enumerated before launch, not after an incident.
- Systems are designed so that when they fail, they fail in ways that are visible and recoverable.
- Any system that could cause meaningful harm when it fails is designed with a human in the loop.
- Performance and reliability claims are measured, not asserted.
Privacy
We treat personal data as a liability to be minimised, not an asset to be accumulated. Concretely:
- We do not collect data we do not need to deliver the service being used.
- We do not build profiles of users across contexts.
- We do not sell data, share it with advertisers, or use it to train models without explicit consent.
- Where we do handle data, we say so clearly — see our Privacy Policy.
Oversight and accountability
Automated decisions that affect people must remain accountable to people. This is not a philosophical position; it is a practical one. Systems that cannot be explained, audited or overridden are systems whose failures cannot be corrected.
Where we build automated decision support, we build it with the expectation that a human will review the output, understand the reasoning, and be able to override it.
Transparency about limitations
We will not overstate the capabilities of the systems we build. When something works, we will describe what it does and under what conditions. When it does not work reliably, we will say so. This is uncomfortable commercially. It is the only position compatible with being trustworthy over the long term.
What this means for you
If you are considering working with us, you should expect these commitments to shape the conversation. We will ask questions about who is affected by a system and what happens when it fails. We may decline work. We will not compromise on the commitments on this page in exchange for revenue.
If you believe we have fallen short of these commitments, please write to us at support@aeraint.online. We take that seriously.