€2.2M · Kazimi + 2 more

What they're building

Kazimi is a security layer that lets a mobile app prove its own telemetry is real. The integration is deliberately small: a developer drops the Kazimi library into the app, the library encrypts signals about the device's execution environment, and the developer decrypts the verdict through an API. What comes back is not a risk score handed down by a third party but a statement about the developer's own first-party data. The detection surface is the classic mobile-tampering set, applied to marketing signals rather than to logins: device emulation, rooted and jailbroken devices, repackaged or tampered builds, code injection, and replay attacks. The company's framing on its own site is that fraudulent signals do not just cost media budget, they poison the downstream systems: bots get treated as high-value users, LiveOps KPIs drift, and user-acquisition spend is allocated against a fiction. The cryptographic choice is the interesting part. Kazimi uses privacy-preserving proofs to establish signal origin and quality without collecting personal data and without asking the user for a permission prompt. That is a specific bet about the post-IDFA world: attribution has already migrated to first-party signals, and nobody has been verifying those signals. The product is platform-agnostic by design, sitting alongside existing MMPs and analytics stacks including Firebase rather than trying to replace them. The named customer shape is mobile gaming studios, rewards platforms, and fintech apps, which is exactly the set of apps where a bot has a direct financial incentive to look like a person.

Why this matters

The founding team built fraud detection at Adjust, the Berlin mobile analytics company AppLovin acquired for roughly $1 billion in 2021. That is not a credential flourish; it is the reason the product exists in this shape. Andreas Naumann ran Adjust's anti-fraud team and drove its bot-detection expansion through the Unbotify acquisition, and his stated principle then was to catch fraudulent activity before attribution takes place rather than reconciling it afterward. Kazimi is that principle rebuilt as infrastructure the developer owns. On timing, the coverage of the round cites a global mobile advertising market of about $419 billion a year and notes that bot traffic is projected to exceed human internet traffic for the first time in 2026. Take the second number as directional rather than gospel, but the direction is the whole argument: as more app signal gets pointed at AI systems and automated bidding, unverified telemetry stops being a media-buying nuisance and becomes corrupted training and optimisation data. The honest counter-argument is that MMPs and ad networks already sell fraud detection, and app teams are tired of buying it. Kazimi's answer is architectural rather than commercial: the verification happens inside the app the developer controls, cryptographically, before anyone else in the chain gets a chance to grade their own homework.