About the PiML integration
PiML is an open-source Python toolbox for interpretable machine learning and model testing, developed under the guidance of model risk management expert Agus Sudjianto.
What you can do
- Bring PiML interpretability and model testing results into Asenion assessments.
- Use the results as evidence for model risk and explainability controls, such as those required under OSFI E-23 and SR 26-2.
Learn more on GitHub.
How it fits together
One set of controls across your whole AI stack
Whatever platforms, models and clouds you use, the Asenion AI Control System applies the same controls to every AI system and agent.
01 · ControlGen
Translates regulations, standards and your policies into control objectives and controls.
02 · Assess
Applies the controls across the AI lifecycle and shows which are met and where the gaps are.
03 · Red-Teaming
Tests AI systems and agents against the controls before deployment.
04 · Witness
Applies and evidences the controls at runtime, with tamper-resistant records.
Get started
See this integration on your own AI systems
In 30 minutes we'll map the rules that apply to your AI, show the controls, and walk through how this integration fits your stack.