Cimento AI

We'll deepfake your CEO. With their sign-off, we clone one approved voice, call the teams you pick, and send you results by role. A blameless coaching moment follows for anyone who was fooled. Free for qualifying security teams.

Three seconds of audio is enough to clone your CFO's voice with an 85% match (McAfee, 2023). Every earnings call and keynote your execs give is source material. We'll run a consented voice-clone drill against your finance and help desk teams and show you who would have acted on it. Free for qualifying security teams.

"The attack surface is not static anymore. We do not have full visibility into everything the agents are doing." That's Derek C. Head of Security at Together AI. It's also why this case study is our favorite one yet. Together AI came to us for security awareness training. Real problems, unglamorous ones: hundreds of manual evidence tickets in a single audit week, onboarding edge cases polluting reports, campaigns that took days to stand up. On Cimento AI: training consolidated across nearly 400 employees, new-hire cohorts automated, content rebranded to look like it comes from Together AI instead of a stock vendor. In Lisa R.'s words, standing up a campaign "went from a manual process with multiple steps to something that can be done within minutes." Then the relationship changed. As agents took on real work across the company, Derek brought us a problem traditional tooling wasn't built for: which agents are running, which employees are behind them, and what can they access? So we're building the answer together. Agent Hub ties every agent's behavior back to a responsible human owner, and our multi-turn simulations now pressure-test the agents themselves, not just the people around them. From training platform to design partner on securing AI. Read the full case study: https://lnkd.in/g6EZcwhk

Every security awareness program is built around the same person: the employee who's behind on training and clicks whatever lands in the inbox. That picture is out of date. The riskiest people in most companies now are the most forward-thinking ones. The engineers adopting agents, approving OAuth scopes, handing automation credentials. Not maliciously. Enthusiastically. A click is discrete and trainable. An allow is continuous; it compounds, and it delegates. Most measurement programs are still pointed at the click. New post on where to point the instrument instead, starting with your strongest people rather than your weakest: https://lnkd.in/g-ue7qbZ

Every vendor wants to show you more signals. It makes the demo look great. It also makes your queue unusable. Our founder Zain Rizavi went on the Talking Cyber Podcast with Kyle McIntyre to talk about the CISO's signal-to-noise problem and why human risk makes it worse: more telemetry, more alerts, and still no answer to "which of my people should I actually worry about today?" Our approach is the opposite of the signal dump. Visibility across your human layer first, then we work with you to understand how your team runs security and who your riskiest users are, so what reaches you is filtered to the risk that's actually moving. More signals isn't the product. Knowing which humans need action is. Full episode: https://lnkd.in/g9_MyE_k

"We're going to get bit by this. We've been bit by this." A head of security said this week about his own annual training program. His team is writing a postmortem on a social engineering incident right now. The training existed. Completion was logged. The attack worked anyway. New on the blog, an uncomfortable argument: checkbox training is worse than no training, because it converts a live risk into a closed ticket. The board sees a completion rate. The attacker sees the same untested people they saw in January. It ends with four moves any security team can make this quarter, starting with this one: pull the completion rate off your board slide and replace it with repeat-failure rate per person. Read it: https://lnkd.in/gh_vynNy