YRIKKA automates red teaming of vision models, building towards a future where humans and AI agents can automatically ensure the reliability of visual AI.
New research from the Yrikka team: Text-to-Model (T2M).
You can now create a custom object detection models with nothing more than a prompt and a few raw images.
No datasets. No labels. No annotation grind.
Deploying vision models?
Crushed the benchmarks, but failing in the wild?
Edge cases, long tail, unexpected inputs: We’re tackling that head-on.
📍 Find us at Booth 1332 at @scsp_ai AI Expo 2025.
Let’s talk about building vision systems that actually hold up.
#SCSPAIExpo25
From drones to robotics to manufacturing -- as AI moves into our physical world it is vital AI results are accurate and can be trusted.
Today, @YRIKKA_Inc launched APEX, a developer-first API for stress-testing visual AI models. An automated red teaming engine - uncovering weaknesses in your specific context, with your models.
Whether you’re deploying computer vision in high-stakes applications or researching robustness and bias, APEX gives you:
- Automated edge-case simulation
- API-based integration (fast and simple)
- Contextual metrics for real-world evaluation
We're blown away by this SMALL BUT MIGHTY team lead by Kia Khezeli and John Kalantari, PhD -- landing Department of Defense contracts and now distributing access of this next generation evaluation tool to all.
Head over to https://t.co/uYDxs8sUHv to try it out.
What if you could red team your object detection models like a pro?
We just launched APEX API – stress-test your visual AI in real-world scenarios via API.
Open access now live https://t.co/bFs93C3SOs
#AI#ML#ComputerVision#DevTools#ProductHunt