"Orca Catches Hidden Exposure from AI Agent-Driven Feature Experiments"
Our product team constantly runs experiments, turning features on and off with flags that are increasingly evaluated or executed by AI agents. Orca doesn't try to live in the feature-flag layer; instead, it watches the underlying cloud identity and the agent posture. When an experiment gives an agent broader access than intended, or opens up a test path to sensitive data, Orca picks it up on its next assessment of the cloud and agent surface and flags it. Developers don't have to change how they ship experiments, but security can still see where agent-driven flags might create unexpected exposure. Explaining to some product folks why the temporary agent-powered experiments still need proper access controls took a few rounds of education, even with Orca's findings on hand. It keeps up with experimentation-heavy, agent-driven development without slowing it down. Orca watches the surfaces where risk actually materializes—when agents toggle, evaluate, or act on feature flags.
"Orca Brings AI Agent Risk into Clear Business Context"
Our finance and enterprise risk teams wanted to understand AI agent risk in business terms, not just issue counts or a vague idea of 'agent exposure.' Orca's context—what data an agent can touch, whether it can reach internet-facing services, and which identities and systems sit behind its paths—lets us group agent-driven risks around business-critical assets and regulated workflows. It's not a full quantitative risk engine, but it has given us a structured way to discuss which agent behaviors matter most to revenue, compliance, and our contractual obligations. We still add some business mapping on our side for certain agent-centric technical risks, so they line up neatly with revenue streams and contractual exposure. It has moved our security discussions with Finance and Enterprise Risk from 'here is a pile of agent findings' to 'here are the top agent-driven threats to the system' that actually drive cost and compliance exposure.
"Orca Clarified Sensitive Data Exposure and AI Agent Access Paths"
Orca gave us a much clearer picture of where sensitive data actually lives in relation to our AI agents. It surfaced PII in storage, secrets in places agents might read, and old database snapshots sitting in accounts that newly created agents could access. The real value is the context: each finding shows what the data is, whether it's exposed, and which agents and identities can access it. That turned what could have been an overwhelming inventory into a clear, prioritized list of data our agents should never be able to touch, along with the paths we needed to close. We spent a bit of time upfront tuning the data classification and agent reachability settings to align with our own definitions of sensitive information, as well as the agent access paths we consider unacceptable. It folds data security into the same agent-aware risk picture as misconfigurations and identities, so 'where is our sensitive data, and which agents can reach it?' is finally a question we can answer continuously.
"Orca Delivers a True Shift-Left Win for IaC and Pipeline Security"
Having Orca scan our infrastructure as code and our pipelines has been a real shift-left win for the AI agents. We started catching templates and configuration snippets that would have given agents broad, long-lived permissions or exposed new agent endpoints before any of those changes deployed. Because Orca ties these IaC findings back to the same risk model it uses for the running environment, we can clearly see that a misconfiguration flagged in code is the same kind of agent exposure we'd otherwise be chasing down in production later. We worked closely with our platform team to integrate agent-focused checks cleanly into our build steps, and to decide which misconfigurations should trigger a warning versus block a deployment. It shifts cloud and agent security left in a way that actually sticks. We're catching reachable, exploitable agent misconfigurations during code review instead of after agents are already live and exposed, which has made the whole pipeline safer without adding friction.
"Orca Surfaced Agent-Reachable Secrets Fast with Clear Context and Minimal Noise"
We had a general feeling that credentials and secrets were scattered around our environment, but we didn't have a clear map of which ones agents could actually access. Orca revealed hardcoded API keys, SSH private keys, and database credentials in places agents or their tools might read—config files, storage buckets, and logs. It doesn't just pattern-match; it uses entropy and the surrounding context to decide if a string is genuinely a secret, which reduces noise. Seeing each secret in the context of agent reachability made it clear which exposures could be triggered by agents and needed urgent cleanup. We spent some time defining suppression rules for test credentials in our sandboxes, so that agent-relevant secrets stayed prominent in the findings. It turned secret sprawl from an abstract concern into a concrete, prioritized cleanup list, with a special focus on secrets that AI agents could stumble over. Exposed credentials are now a tracked, agent-aware risk rather than something we'd get to eventually.
"Orca Found Hidden Malware and Suspicious Workflows Our Other Tools Missed"
Orca flagged malicious files and suspicious workflows across our workloads that our other tools hadn't detected, including cases where an abused AI agent could have been used to plant or trigger malware. It goes beyond simple hash matching, catching things like cryptominers and web shells that agents might deploy or invoke as part of a compromised toolchain. Because scanning happens off the workload at the storage and configuration layer, there's nothing for malware to hide from on the box itself, and it also covers hosts and agent-linked paths we never would have pointed a dedicated scanner at. We aligned Orca's agent-aware malware and threat findings with our existing endpoint response processes, which helped us fold that context smoothly into how the team already works. It gave us malware and abuse visibility across our cloud workloads and the agent paths that were previously blind spots. We can see not just that malware exists, but how AI agents could be used to deploy or spread it, and then cut those routes off.
"Orca's Attack Path View Revolutionized How We Prioritize Fixes"
The game-changer for us is the attack path view. Instead of a list of isolated findings, Orca shows the complete chain: an exposed asset, the vulnerability on it, the identity it can assume, and the sensitive data at the end. Seeing the entire route makes it obvious which single fix can break the chain, so we stopped trying to patch everything and focused on cutting the links that matter most. In a large environment, the attack path graphs can be dense, so we found it best to scope to our crown jewel assets first. Once we did that, the overall picture became clear and useful. It moved us from just counting vulnerabilities to reasoning about exploitable routes, which is a much better use of a small team's time. It's especially valuable when AI agents are part of the route, because we can see how agents access identities and how data paths fit into the real chain an attacker could follow. As a result, we fix fewer things, but we reduce more actual risk.
"Orca Provided Fast Visibility into AI Agent Risks"
Our company operates at high speed. Developers, analysts, and PMs are constantly releasing, and AI agents are embedded in many of those processes. We needed visibility that could match that momentum—not a sluggish rollout that would still be unfinished by the time the next agent launched.
Orca connected to our cloud account, and within days we had a real risk picture, including the AI agents: the identities they operate under, the endpoints they expose, and the data they can reach. We didn't have to chase teams to get a clear view of our agent services. The main follow-up was organizing the findings to reflect our team and business unit structure, including grouping agents by owner and function. That organizational step has made ongoing reviews much clearer for leadership. This removed the typical deployment obstacle and gave us AI agent visibility before we could even finish planning the rollout. That speed matters even more as agents and their workflows spread across the state, because they become visible and manageable as quickly as teams create them.
"Orca's Exposure Context Makes Prioritizing Host Vulnerabilities Simple"
Everyone talks about containers these days, but we still have plenty of VMs and long-running hosts, and our old scanning was weakest there. Orca treats them the same as everything else, showing which host vulnerabilities are on internet-facing machines, which are connected to privileged identities, and which are actually reachable versus protected by existing controls. That extra context lets us prioritize host patching based on real exposure rather than just CVSS scores. Some of the older hosts had the usual configurations that needed a second look, but the findings were solid, and the added context made them easy to understand. It filled the gap between modern container coverage and our older VM estate. Now host risk is prioritized based on actual exposure, not just severity numbers. It also helps us see when host and VM risk might affect the infrastructure supporting AI agents or the systems those agents rely on.
"Orca Exposes Shadow Data Risks with Straightforward Sensitive Data Discovery"
We kept running into the issue of production data being duplicated into staging and development environments that weren't as tightly controlled, along with storage we didn't even recall setting up. Orca identifies sensitive information across managed, unmanaged, and shadow repositories, including files on VMs, containers, and buckets, and sorts it into groups such as PII and payment details. Spotting a forgotten production snapshot in a loosely managed dev account and seeing how that connection highlighted the account's exposure made it clear which copies were actually dangerous versus harmless. We had to fine-tune things a bit to separate intentional copies from risky ones, since some lower-environment data was supposed to be there. Still, setting it up to fit our workflows was quick. It offers genuine insight into sensitive data sprawl, including shadow and misplaced data, so the divide between secure production and less-protected accounts no longer hides as an invisible threat. It also points out where AI agents might stumble onto sensitive data they shouldn't access, turning those risky paths into concrete action items.