AI security
We test models, agents and RAG to see whether they can be tricked, reveal information or use tools beyond what was intended.
Lucis / APPLIED SECURITY
We work with teams building digital products, using AI or moving sensitive data. We test what can fail and turn it into concrete decisions.
Tell us what you are buildingFOR TEAMS THAT BUILD
We translate technical problems into something the whole organization can understand: what could happen, why it matters and what comes next. Offensive thinking, business judgment and work that can be put into practice.
EXPERIENCE
The best signal of security work is what it enables next: more confident AI adoption, better architecture decisions and a repeatable standard for every client.
The team is helping us integrate AI safely, not only into our integrations and the platforms we use, but also into the way we bring AI into our own workflows.
A technical reference of enormous value, both for the depth of his knowledge and for his ability to apply it to real problems. He made a key contribution to our AI architecture initiatives and helped us define an implementation model we now repeat with our clients.
WHAT WE DO
We understand the whole path: people, product, APIs, cloud, models and data. That is how we find attack paths that make sense in the real world.
We test models, agents and RAG to see whether they can be tricked, reveal information or use tools beyond what was intended.
We check whether an account can see or do things it should not, change a business rule or chain small weaknesses into a real path.
We review who can enter your cloud, what they can do and how to bring security controls into development without slowing the team down.
One complete program for phishing, SMS, web fake and deepfake simulations, followed by videos and resources that turn each measurement into habits the team can sustain.
Explore the full serviceHOW WE WORK
We fit into your team’s pace. The result is not a PDF to file away: it is clear evidence, shared priorities and changes that can be implemented and checked again.
We map the architecture, assets, data flows and trust boundaries.
The whole system before the tool.
We look for attack paths across the agreed surface, with authorization and a scope everyone can sustain.
Evidence should be repeatable.
We sort findings by impact and effort so the team knows what to move first.
Every finding ends in a decision.
We retest and leave controls that make sense for the next cycle.
Security stays in motion.
START HERE
Tell us what you are building, what already worries you and what you need to decide. We will come back with an honest first view.
hello@lucis.pro