Cloud security teams are being pushed toward natural-language tools as environments grow more complex. Inventories are bigger, relationships are harder to track, and investigations cut across more data sources. But many AI-driven tools still behave like black boxes. They return answers without showing how those answers were formed, which makes teams hesitate to rely on them during real incidents.
That trust gap is what
Upwind is targeting with the launch of Choppy AI. The goal is not just faster searches or friendlier interfaces, but AI that exposes its reasoning and stays grounded in real cloud context.
What Choppy AI does differently
Upwind positions Choppy AI as more than a generic natural-language layer.
Amiram Shachar, CEO and cofounder of Upwind, told ChannelE2E, “Choppy AI is a dedicated agent trained on specific security datasets and each customer’s unique context. Using natural language, it delivers simple, accurate, and fast answers.”
Those answers are designed to be explainable by default. “Every answer not only provides a clear conclusion, but also supports it with charts, tabular data, and aggregated references from across the platform, ensuring ease of use and full transparency,” Shachar says. The intent is to let teams see not just the outcome, but the evidence behind it.
Visible queries and rules, not hidden logic
A core part of Choppy AI is how it handles queries and policies. Natural-language prompts are translated into structured logic that remains fully visible and editable inside the platform. Teams can inspect how conditions are built, reuse them, or adjust them if needed.
In practice, Upwind is aiming for minimal adjustment. “Our goal is 100% ‘accept it as-is,’” Shachar says. “This is closely monitored by our product teams so that if customers find Choppy AI’s answers inaccurate, we can continuously improve the model.” Transparency here is not about encouraging constant edits, but about earning confidence over time.
From search to conversation
Upwind expects Choppy AI to gradually change how security teams work day to day. Instead of navigating menus or building searches manually, teams can investigate issues through conversational interaction that stays tied to runtime data and asset relationships.
“We believe that, much like the ChatGPT experience, customers will gradually shift from performing daily tasks through traditional search to engaging in GPT-style conversations, eventually transforming how they work within the Upwind platform over time,” Shachar says. That shift has implications for how policies are created, how investigations unfold, and how risk is prioritized.
What this means for MSSPs and partners
For MSSPs, Choppy AI is positioned as an embedded capability rather than a separate product. “Choppy is embedded across the entire CNAPP platform, and MSSPs using Upwind can benefit from a focused, context-aware agent across all managed accounts,” Shachar explains.
Just as importantly, it does not introduce new commercial complexity. “It does not change pricing or packaging,” he says. That makes Choppy AI easier to absorb into managed CNAPP and cloud security services without reworking contracts or service tiers.
Choppy AI reflects a more restrained approach to AI in cloud security. Instead of promising autonomous decisions, Upwind is emphasizing visibility, context, and control. For teams wary of opaque automation, the message is straightforward: AI can speed up investigations and policy work, but only if teams can see and trust how it reaches its conclusions.
