Upwind announced Choppy AI, embedding new AI-powered capabilities across the company’s CNAPP platform. Choppy AI introduces natural-language–driven experiences that make cloud security exploration, investigation, and analysis more intuitive, while providing transparency, control, and trust for security teams.
As cloud environments grow more complex, security teams are required to manage large inventories, relationships, configurations, and vulnerabilities. While AI has the potential to simplify this work, most AI-driven security tools operate as black boxes, producing answers without exposing the logic behind them. As a result, Upwind has launched Choppy AI to take a different approach.
“Choppy AI is designed to make cloud security exploration and investigation faster and more intuitive while making the logic behind it clear and auditable,” said Amiram Shachar, CEO at Upwind. “By translating natural-language intent into structured, editable queries and rules, we’re giving teams the speed of AI with the transparency and control they need to trust the outcome. Every AI-generated result is visible, editable, and enforceable. Teams can see exactly how logic is constructed and apply it with confidence in production environments.”
Transparent, controllable AI built for cloud security work
Choppy AI translates natural-language intent into structured, visible logic that security teams can inspect, edit, and reuse across the platform. AI-generated outputs remain transparent and enforceable, ensuring teams retain ownership over how decisions are made.
Rather than operating as a black box or replacing existing workflows, Choppy AI augments them, accelerating analysis while grounding every result in the same runtime and contextual data that powers Upwind’s CNAPP.
Natural-language inventory exploration, rule creation, and investigation
Choppy AI enables security teams to explore cloud assets, define security logic, and investigate risk using natural language across the platform. Users can search inventories using free-text input, which Choppy AI converts into customizable Query Builder expressions, making it easy to express complex intent while seeing exactly how each condition is defined.
For security policy creation, teams can describe configuration or relationship-based logic in plain language, which Choppy AI translates into transparent, editable rules used to detect misconfigurations and evaluate exposure paths without manually building queries or graph logic.
In the Vulnerability module, Choppy AI introduces an interactive AI Mode that enables conversational investigation. Security teams can ask questions, follow up iteratively, and explore vulnerabilities directly within the platform, with responses grounded in real assets, relationships, and runtime exposure to help prioritize what truly matters.
Designed for trust, monitoring, and improvement
Each Choppy AI capability is supported by dedicated monitoring to track real-world usage patterns, prompts, and behavior. This allows Upwind to continuously refine the experience while ensuring AI behavior remains observable, predictable, and aligned with security team expectations.
Choppy AI is generally available to all Upwind customers and represents a major step forward in delivering practical, trustworthy AI that accelerates security workflows without compromising control.
The launch of Choppy AI builds on Upwind’s recent introduction of Inside-Out AI Security across its unified CNAPP platform.
