Get ready for the cyberattack. Train your team to restore critical operations safely, with AI as support.
Hands-on programmes for security, IT and incident response teams. You build and test AI-supported workflows, rehearse incident handling and practise safe recovery in a contained exercise environment.
For financial services, energy and utilities.
When the alarm goes off on Friday morning
How quickly can you gather the evidence, understand the scope and decide on the right interventions yourselves? Attackers have automated their bottlenecks with AI agents. In many organisations, these tools are not yet part of incident readiness. And every agent the wider organisation deploys is a new identity with access that the security team must govern.
29 minutes
Average time from initial access to lateral movement in the network.
CrowdStrike 2026 Global Threat Report24 February 2026
About minus 7 days
Average time from public disclosure of a vulnerability to exploitation, often before a patch is available.
Mandiant M-Trends 202623 March 2026
High
On 24 September 2026, the Danish Agency for Societal Security raised the threat level for destructive cyberattacks against Denmark from medium to high.
Danish Agency for Societal Security24 September 2026
Threat report figures are reproduced as reported by their publishers. Status as of October 2026.
The programme follows your incident readiness
Build
AI support for the first hours of an incident. A faster overview, a timeline where every claim can be checked, and suggested interventions that humans approve.
Protect
The agent is itself an attack surface. The team learns to build agents that are harder to misuse, for example through manipulated log content.
Govern
Agents are identities with an owner, permissions and a lifecycle. Clear rules define who may decide on interventions.
The agent sees logins at 03:12 on a service account from an unknown IP range and concludes that this is planned maintenance because a field in the log says so. The attacker wrote that field. In the exercise, the team learns to catch the error with two checks: text in log fields is never an explanation on its own, and maintenance must match a registered change. No change, no clearance.
Exercise scenarios for your sector
The scenario is selected during preparation and uses synthetic data that resembles yours. Examples:
Financial services
- Outage of a critical customer-facing service, such as online banking or payments.
- Compromised identity with access to customer data.
- Outage or compromise at a critical ICT third-party provider.
- Classification and reporting of a major incident.
Energy and utilities
- Ransomware in IT with a risk of spreading towards operations.
- Safe recovery of administrative systems while essential services keep running.
- Communication to customers and authorities during service disruption.
Exercises never change industrial control systems or production environments. Agents have read access, and authorised humans decide on all interventions.
- SOC: alert triage and enrichment, shift handover summaries.
- Identity and IAM: access reviews, access requests and separation of permissions.
- PAM: privileged sessions and anomalies in emergency access.
- Incident response: timelines, communication and incident handling support.
- Vulnerability management: assessment and prioritisation of vulnerabilities.
- GRC: policies, control requirements and evidence.
The framework is already in place
Financial services: DORA has applied since 17 January 2025. It requires response and recovery plans, backup and restoration, readiness testing and training in digital operational resilience, including for management.
Energy, utilities and other essential and important entities: the Danish NIS2 Act took effect on 1 July 2025. It makes cyber security a management responsibility and requires training. Some sectors have additional rules.
Since 2 February 2025, AI Act Article 4 has required organisations to ensure, to their best extent, sufficient AI literacy among staff working with AI.
The exercises provide documented training and tested workflows you can use in your own work with these requirements. You decide whether and how they fit into your programme together with compliance and risk functions.
General guidance, not legal advice.
Who is it for?
- SOC and monitoring.
- Identity, IAM and PAM.
- Incident response and vulnerability management.
- GRC, compliance and the CISO office.
- IT operations and platform teams.
- Incident readiness and business continuity owners.
- ICT risk and second-line functions in financial services.
- Operations owners in energy and utilities.
Participants need to know their own work. Coding is not a prerequisite.
The three packages
Up to 20 participants in each package. Exercises use synthetic data in an isolated environment.
Kickstart: try AI in your cyber readiness
½ day · 3½ hours
For teams that want to understand the possibilities and choose a concrete starting point.
- Focus
- A shared security case using synthetic data. The team builds and tests a first agent and sees local AI as a possible part of incident readiness.
- Deliverables
- A tested training prototype, a completed Agent Canvas, three prioritised applications and a recommended next step.
- Local AI
- Possible part of incident readiness
Hands-on day: train one critical workflow
1 day
For teams that want to work thoroughly on one contained task in their incident readiness.
- Focus
- One critical workflow from your incident readiness, such as incident analysis or recovery planning. Built and tested under realistic time pressure with clear human decision points.
- Deliverables
- A tested prototype for the chosen workflow, a documented workflow with roles and checkpoints, a list of identified gaps and a 30-day action plan.
- Local AI
- Agreed during preparation
Bootcamp: practise safe recovery
2 days
For teams that want to rehearse a connected incident and recovery sequence.
- Focus
- A connected incident and recovery sequence in an isolated exercise environment with local AI support. The team rehearses from the first alarm to a restored service, then repeats a new exercise with less help.
- Deliverables
- A tested exercise sequence and recovery guide, documented setup of the local AI support, a short exercise report with measurements and prioritised improvements, and a 30-day action plan.
- Local AI
- Local AI support in an isolated environment
Platform, hardware, installation permissions and exercise scope are agreed and tested before the programme. Offline exercises require models, installation files and dependencies to be available locally.
| The three packages | Kickstart | Hands-on day | Bootcamp |
|---|---|---|---|
| Duration | ½ day · 3½ hours | 1 day | 2 days |
| For whom | Teams choosing a starting point | Teams with one contained task | Teams rehearsing safe recovery |
| Focus | Shared security case | One critical workflow | Incident and safe recovery |
| Deliverables | Prototype, Agent Canvas and next step | Prototype, workflow and action plan | Guide, exercise report and action plan |
| Local AI | Possible part of readiness | Agreed during preparation | Local support in an isolated environment |
| Follow-up | Shared online follow-up | Shared online follow-up | Shared online follow-up |
Price agreed in a short call about the package and team size.
Programmes can run in Danish or English, on-site or hybrid.
What we measure in the exercise
Safe recovery
How quickly an agreed critical function works again and passes the checks.
Team response capability
How quickly the team reaches a verified overview and makes decisions.
Local AI support
Whether AI support can start and complete the task under the exercise conditions.
Reuse
Whether a colleague can reproduce the setup and the team can handle a new scenario independently.
Results apply to the agreed exercise and its conditions. They document team performance and improvement needs. They do not guarantee recovery times in production.
Readiness check: six decisions before the alarm
Agents cannot be introduced into incident readiness in the middle of an incident. Most of the work happens beforehand.
- 01
Which model may see your logs: a local open-weight model, an approved EU-hosted model or your security platform.
- 02
What the agent can access: read access to contained sources, traceable tool use and a fixed output format.
- 03
Who may decide on interventions: disable accounts, isolate servers or close integrations.
- 04
An exercise environment with known answers: synthetic logs with noise, false leads and missing data.
- 05
Can the environment be reproduced: version-locked dependencies and a guide a colleague can follow.
- 06
What you measure: time to a verified basis for decisions, analysis quality and errors caught.
If you cannot answer all six, you know what the first exercise should address.
Bring the six questions to a callPreparation, follow-up and next steps
Every package includes one needs and scope call, selection of a case within the package scope, synthetic data and testing of the agreed tools, materials for the format and one shared 45-minute online follow-up around 30 days later to review use, barriers and next steps.
Optional by agreement: separate development, production integrations, further advisory and support.
Keep your readiness rehearsed
Optional quarterly exercises, purchased separately: a new contained scenario, retesting of workflows and local AI support, comparison with earlier results where exercises are comparable, and updates to guides and improvement plans.
- Agent Governance Sprint.
- Advisory on Agent 365 and Entra Agent ID.
- Leadership tabletop.
- Leadership briefing.
What participants say
"Stefanos fagligt høje kompetencer. Hans passion skinner tydeligt igennem, og han gør et svært emne spiseligt. Jeg føler mig mere oplyst og sikker i brugen af AI."
Deltager, Lær at bruge AI-agenter (ITU, sep. 2026)
"Den røde tråd, hvor man hele tiden bygger nye lag på sin løsning. Jeg har fået nye idéer til, hvordan jeg kan videreudvikle mine skills."
Sune Mehlsen, Lær at bruge AI-agenter (ITU, sep. 2026)
"Strukturen og metoden, tempoet og underviserens evner i formidling. Jeg lærte at bygge agenter, og logikken og forståelsen bagved."
Pia-Marie Schougaard, Lær at bruge AI-agenter (ITU, sep. 2026)
"Kæmpe øjenåbner. Kurset var meget inspirerende og lærerigt med masser af hands-on læring, og jeg vil helt sikkert bruge læringen fremadrettet."
Gitte Friis-Elkjær, Digital Produktudvikling med AI-agenter (ITU, jun. 2026)
"Det praktiske element, at bygge noget on the spot, har været rigtig godt. Use-cases var rigtig gode og tilpasset på en relevant måde. Jeg kan bygge noget brugbart selv og se en vej til at lære mere på egen hånd."
Andreas, Digital Produktudvikling med AI-agenter (ITU, jun. 2026)
"Underviserens entusiasme, og jeg fik sat en del begreber på plads. Jeg fik redskaber til at kunne gå i gang med det samme på arbejdspladsen."
Nynne Vidgren, Digital Produktudvikling med AI-agenter (ITU, aug. 2026)
"Vi kom igennem et konkret hands-on forløb, hvor vi fik prøvet forskellige programmer af. Jeg lærte, hvordan jeg får sat et projekt og en agent op, som kan hjælpe mig i mit daglige UX-arbejde."
Janne Rounborg, Digital Produktudvikling med AI-agenter (ITU, aug. 2026)
"Hands-on og vekslen mellem hands-on og undervisning, og en super godt forberedt instruktør, der også var super inspirerende."
Allan Orup Hansen, Lær at bruge AI-agenter (ITU, sep. 2026)
"Praktiske øvelser med AI-agenter. Jeg lærte, hvordan man giver en AI-agent tydelige instruktioner."
Reza Neshatfar, Lær at bruge AI-agenter (ITU, sep. 2026)
"Der var tid til at få erfaring med de forskellige værktøjer. Jeg lærte at bruge Agent Canvas og at bruge de forskellige modeller til at forbedre hinanden."
Ane Brøndberg, Digital Produktudvikling med AI-agenter (ITU, aug. 2026)
"Jeg lærte om nye tools, jeg ikke havde prøvet før: Lovable og NotebookLM. Og hvordan man splitter store komplekse projekter op og adresserer dem."
Kell Michael Jensen, Digital Produktudvikling med AI-agenter (ITU, jun. 2026)
"Rigtig god introduktion til AI-værktøjer og gode hands-on opgaver. Jeg har fået struktur i mine prompts og lært at bruge meta-prompts."
Rasmus Gaardsøe, Digital Produktudvikling med AI-agenter (ITU, jun. 2026)
"Engageret underviser og god struktur. Jeg lærte at oprette agenter med canvas, guardrails og menneskelig validering."
Deltager, Digital Produktudvikling med AI-agenter (ITU, aug. 2026)
"Stefano's infectious enthusiasm brought the teaching to life. The hands-on approach worked perfectly; I got to try many AI tools and now feel ready to use them in practice."
"Good balance between presentations and exercises. Stefano is dynamic and engaged, and his deep knowledge of AI products clearly shows in the teaching."
"Really good and exciting exercises with relevant real-world cases that showed how AI is actually used in other Danish companies. Super inspiring!"
"Stefano's huge preparation work shines through. Plenty of material and hints for further study – I have resources to dive deeper on my own."
"The ethics debate was an important and welcome element. The course isn't just about tools, but also about responsible use of AI in practice."
"Learning how to learn with AI was a gamechanger. Real world examples and best practices mean I can now continue my development on my own."
Your trainer
Stefano Vincenti has worked on business-critical IT in banking and payment infrastructure. At Saxo Bank he spent three years helping build the bank's Cloud Factory from scratch: a secure and compliant hybrid and multi-cloud platform, developed in close dialogue with the Danish FSA up to board approval of cloud outsourcing. At Nexi Group he was Release Train Engineer for the teams behind MitID, Denmark's national digital identity with more than five million users, and led releases including channel binding against social engineering. At BEC he managed the renewal of business-critical central bank systems with 24/7 operations and led API and security infrastructure for digital banking. He has also worked in critical infrastructure at TDC NET and DSB.
In June 2017 he was on site at a large Danish shipping company when NotPetya hit, and watched the computers shut down one by one. That is why he believes readiness must be rehearsed before it is needed.
Today he works as a Fractional Chief AI Officer, teaches AI agents as an external lecturer at the IT University of Copenhagen and DIS, and works hands-on as an administrator in Microsoft Agent 365, Power Apps and Copilot Studio.
Frequently asked questions
Ready to rehearse your cyber readiness?
Talk to us about your team, critical operations and the package that fits.
Book a call about your cyber readiness