Meeting privacy research · Ankara, Türkiye
The idea we are researching: environments where a confidential conversation cannot be quietly recorded. The hardware is a concept, not a product. We also build software with security designed in from the first line.
Claude and other tools help us work carefully. A person reviews every change that matters.
How a change ships
TodayA change is proposed
Small, reviewable and described.
Claude reviews it
Injection, broken access control, leaked secrets.
Tests cover abuse cases
Not just the happy path.
A person approves
Nothing is merged or deployed without it.
Secure by design
Security decisions are made when the architecture is drawn, not after launch.
Human in the loop
AI suggests, people decide. Every change is reviewed before it ships.
Least privilege
Agents and services get only the access their task requires.
Precise about our stage
We are early-stage R&D, and we never claim a product before it exists.
Research
Our first phase is R&D. The lead idea is secret meetings in rooms that resist quiet recording. Nothing below is a shipping product yet. Each area is labelled with its current stage.
A portable device concept for secret, in-person meetings. It is meant to help protect the room from covert audio recording, so what is said there stays in the room.
Early concept visualisation (AI-generated illustration, not a photo of a finished product).
Web applications and services where authentication, authorization, input handling and secrets management are designed in from day one, not bolted on later.
A repeatable process for small teams: AI-assisted code review, threat modeling and security tests, combined with conventional static analysis and hardened deployment baselines.
AI, with care
We are building our own security technology and want it secure from the start. Claude is one of several AI and engineering tools we use along the way, and we plan to rely on it more as our tooling matures. AI helps us do the work. It is not the product.
Every suggestion is reviewed by a person before it is merged or deployed.
We try to break our own products before anyone else does. That means adversarial and abuse-case testing, fuzzing, stress and load tests, failure injection, red-team style reviews and long-running endurance tests. Claude and other tools help us generate test cases, review code and widen coverage. A person reviews and signs off every change.
Where we are
We prefer to be precise about where we are.
Team
SpySec is led by its founder, with a background in software development and security.
Emre Erdoğan
Founder & CEO
Contact
We are happy to hear from people working on secure software and privacy, potential partners and early users.
contact@spysec.eu