Connects to what you already have.
Works alongside your EDR, your firewall and every source you connect; it doesn't replace them.
AI Technologies
Cybersecurity ecosystem · Made in Chile
Cisent tests your defenses, contains attacks in real time, documents every incident and trains your team. Everything runs on AI models trained by Cisent.
Attacks no longer arrive only in writing: Cisent reads, sees and listens.
Programs
JumpStartHow it works
The four products feed each other inside Husker, our own agentic layer. Red Team finds the gap, Defense blocks it, Report keeps the evidence and Cisent Iroh turns every case into learning for your team.
The foundation, in three layers
Features
Works alongside your EDR, your firewall and every source you connect; it doesn't replace them.
Not even anonymized.
Cisent trains and operates them; how they work doesn't depend on foreign providers' usage policies.
They act on their own where an action can be undone and ask before anything with impact. If an action can't be recorded, it doesn't run.
The models flag doubt instead of making up a fact: Mind Ultra 2 fabricates 60% less than its base.
Development, operations and support in Chile, in Spanish, with local regulation as the starting point.
Products
Defense, Red Team, Report and Cisent Iroh share models, agents and evidence. What Red Team uncovers, Defense blocks; what Defense contains, Report documents.
Sentinel Defense
Detects, assesses and blocks attacks alongside your EDR, your firewall and every source you connect. It acts on its own only where an action can be undone, and asks before anything with impact.
Sentinel Red Team
AI agents attack your systems with your authorization and within the scope you define, to find the gaps before an attacker does.
Sentinel Report
Gathers and seals the evidence and drafts the report for each audience. For any company, whether or not Law 21.663 applies to it.
Cisent Iroh
Talks with your team, explains alerts and incidents and teaches how to respond. It's an add-on to any of the other three products.
Manthys
Manthys is four families of models. Every figure carries its source.
Manthys Mind · generalist
60%
less fabrication than its base.
Mind Ultra 2 on SimpleQA: 36.7% incorrect answers vs. 90.9% for its base.
Manthys Sentinel · cybersecurity
72.44F1
on ATT&CK techniques; the best figure published by another specialist is 49.1.
Sentinel Ultra 2 on CTI-Bench (CTI-ATE, 60 reports). 49.1: Cisco's Foundation-Sec-8B-Reasoning (arXiv 2601.21051).
Manthys Nova · the largest scale
22.7%
of its reasoning traces state uncertainty.
Nova C 2, CTI-MCQ reasoning traces (150 questions).
Manthys Reflex · decisions
Reflex C 2 reads every log and answers with a decision and its probability; it doesn't write.
27of 28
injected attacks detected in 48 hours of real traffic, measured blind.
A lab firewall, 7,401 windows. False alarms per day: 3 on “port scan”, 5 on “alert the judge” and 0 on brute force.
Explore ReflexScience and research
Getting started
We assess your current defenses and hand you the vulnerabilities we find, before implementing anything else.
Sentinel Defense connects to your firewall, your EDR and every source you have, with the action rules your team approves.
Sentinel Report is ready to document and report the next attack.
The first days
Team
Cisent in the real world
An alert comes in and it's unclear whether it's an attack. Sentinel Defense gathers logs, traffic and captures; Manthys Sentinel reads them like an analyst and maps the technique in ATT&CK; Husker acts according to the impact.
Sentinel Defense · Manthys Sentinel
Explore Sentinel DefenseA machine starts encrypting files. Defense cuts the intrusive connection, blocks the attacking IP and freezes the evidence on the spot; before isolating a machine in your organization, it asks. Report documents in parallel.
Sentinel Defense · Sentinel Report
Explore Sentinel DefenseIf the law applies to you, the early warning to ANCI is due within 3 hours; the board and the insurer also want their report. Sentinel Report gathers the sealed evidence and drafts each version; your team reviews and sends.
Sentinel Report · Manthys Sentinel
Explore Sentinel ReportRed Team attacks with your authorization and within the agreed scope. What it finds becomes Defense rules and is documented by Report.
Sentinel Red Team
Explore Sentinel Red TeamCisent Iroh explains what happened, which technique was used and how it was handled, in plain language.
Cisent Iroh
Explore Cisent IrohQuestions
Products contracted as a service: Sentinel Defense, Sentinel Red Team, Sentinel Report and Cisent Iroh. The Manthys models are not sold separately.
Yes. Report documents every attack for the board, the team and the insurer; if the law applies to you, it also prepares the reports to ANCI.
No. It works alongside your EDR, your firewall, every source you connect and your team, and asks before any action with impact.
FortiGate, pfSense, OPNsense, UniFi and more, through connectors audited by Cisent.
What can be undone: blocking an external attacking IP, cutting an intrusive connection and freezing the evidence. Blocking an internal machine or changing a rule, it asks first.
Manthys judges: the models analyze and classify. Husker decides: it acts within the approved limits and records everything.
No. Our own models, with no outside AI provider in between. We don't train on your data. Not even anonymized.
No. We don't train on your data. Not even anonymized.
Those set by the Law 21.663 regulation: early warning within 3 hours, second report within 72 hours (24 if an essential service is affected), action plan within 7 days and final report within 15 days.
Blog & news
Threat research
PHP released patches for its four supported branches, with seven critical flaws that require uncommon configurations; two close cases left open by the PHP2Shell fix.
Artificial intelligence
California recommends mandatory standards for AI companies, independent testing, emergency shutdown by state order and a cap on AI that designs its own successor.
Threat research
CISA added to its catalog of exploited flaws a vulnerability in LiteLLM, the open-source AI gateway, that allows code execution on the server and has had a patch since May.
Tell us what your organization needs to protect. We'll reply on WhatsApp and set up the meeting.
Message us on WhatsApp and let's book a meetingor write to info@cisent.ai