1. Platform
  2. Adaptive Risk Campaigns

One adaptive campaign. Every employee gets their own journey.

Set the goal and the guardrails once. Cyberfrog then adapts simulations, lessons and reinforcement to each person’s role, language, skills and behaviour, continuously, and shows you exactly why it did what it did.

Book a demo See three journeys
  • Individual paths, one campaign
  • Driven by real behaviour
  • Explainable risk
  • Autopilot or your own rules
ONE CAMPAIGN · EVERY EMPLOYEE GETS THEIR OWN PATHAdaptive Risk CampaignReduce payment-fraud riskAnna, Finance, DEInvoice simulationClicked, submittedMicro-lessonRetest in 2 weeksBen, Engineering, UKDeveloper noticeReported itPoints earnedHarder scenarioChloé, HR, FRGDPR lesson, FrenchNot completedFriendly reminderCompleted
01Personal journeys

Three employees. Three different journeys.

One campaign adapts its content, simulation type, language, difficulty and follow-up to each person, based on what they actually do.

  1. Anna · Finance · GermanyFalls for an invoice scam

    Gets an invoice simulation in German and enters her password. An immediate micro-lesson on credential phishing follows; when she fails the quiz, different material comes later, and a follow-up simulation checks that it stuck.

  2. Ben · Engineering · UKSpots it and reports it

    Reports a fake developer-platform notice with the report button. He earns recognition in the portal, his risk goes down, and his next simulation is a more sophisticated developer scenario.

  3. Chloé · HR · FranceMisses a deadline

    Gets GDPR training in French, misses the reminder date and receives a friendly, factual reminder. Her compliance status shows overdue until she completes it, then her journey simply continues.

02Setup

Goals and guardrails, not flowcharts.

You define what you want to achieve and what Cyberfrog may do. You do not draw every possible path for every employee: Cyberfrog builds and runs them.

dashboard.cyberfrogsecurity.com/campaigns/new?type=adaptiveAdaptive Risk CampaignSet goals and guardrails. Cyberfrog builds every journey.1ObjectiveReduce payment-fraud risk2AudienceWhole company, 412 people3Starting pointSimulation first4AdaptationAutopilot5GuardrailsMax 2 simulations a month, working hours6ReviewStart, no end dateContinuous enrolment: new employees join automatically,and role or language changes shape what comes next.
  1. Objective

    Reduce human risk, improve phishing resilience, meet a compliance goal or strengthen one skill, such as payment-fraud detection.

  2. Audience

    The whole organisation or selected teams. Department, role, country and language are picked up automatically.

  3. Starting point

    Training first, simulation first, or a baseline assessment.

  4. Adaptation

    Autopilot, or your own rules for follow-ups, difficulty, remediation and reminders.

  5. Guardrails

    Frequency caps, working hours, escalation permissions and exclusions, so nobody is overloaded.

  6. Review and start

    Approve the programme. It can run without an end date: new employees join automatically, and changes of role or language shape what comes next.

03Signals

Every meaningful event becomes a signal.

Cyberfrog records what people do across simulations, learning, policies and reporting, and decides whether a new action is useful, or whether doing nothing is better.

  • Clicked
  • Submitted
  • Reported
  • Avoided
  • Quiz passed or failed
  • Policy acknowledged
  • Deadline missed
  • New role or language
Trustworthy
Only attributable behaviour counts: clicks from mail scanners and email opens do not raise anyone’s risk.
Fair
A missed deadline is evidence about compliance, not proof that someone falls for phishing, and is treated that way.
Later
Identity, DLP and SOC signals can be added through integrations, where events can be attributed reliably.
WHAT CYBERFROG OBSERVES, AND WHAT IT DOES ABOUT ITSIGNALEXAMPLESRESPONSEAttack simulationsClicked, submitted, replied, reported, avoidedAdjust difficulty, micro-lesson, retestE-learningStarted, completed, quiz passed or failedRemind, reinforce weak topics, retestPoliciesRead, acknowledged, deadline missedReminder, reassignment, compliance statusPhishing reportsSimulation or real threat correctly reportedReward and raise the barOrganisation changesNew hire, new role, new language, leaverAdapt content, language and eligibility
HUMAN RISK, IN THREE DIMENSIONSDIMENSION 01BehaviourWhat people actually did in simulations55/ 100, higher is riskierDIMENSION 02Exposure and impactRole, privileges, sensitive data72/ 100, higher is more exposedDIMENSION 03Learning and complianceTraining, knowledge, policies86%, higher is betterDifficulty-weightedRecent counts morePatterns over luckReports count positivelyOnly trustworthy eventsCompletion is not competenceAlways explainedInsufficient evidence shownA deterministic, versioned model. AI may suggest and explain, never change a score on its own.
04Explainable risk

Not a points tally. A credible risk picture.

Human risk is not calculated by adding penalties and subtracting courses. Cyberfrog keeps behaviour, exposure and learning apart, and shows every component with its evidence.

Behaviour
What people actually did in simulations and assessments, weighted by difficulty and how recent it is.
Exposure
How much damage a successful attack on this role could do: privileges, sensitive data, threats it faces.
Learning
Training, demonstrated knowledge, policy acknowledgements and deadlines.
Honest
New employees without enough evidence show “insufficient evidence”, not a reassuring low score.
  1. One profile

    One risk history per person across every campaign, never contradictory scores.

  2. Versioned

    A deterministic, configurable model; every change of the model is versioned.

  3. Explained

    Which events, which weights, which version moved the score.

  4. Not bought

    Portal points reward good behaviour, but never buy a lower risk score.

05Risk automation

The right next step, automatically.

Cyberfrog picks useful interventions, not an endless stream of reminders. Start with Autopilot’s tested defaults, add your own rules, or target a group by hand at any moment.

Failed a simulation
An immediate micro-lesson and a later retest on the same skill.
Failed a quiz twice
Different material for the same weakness.
Reported three in a row
Higher difficulty within your limits; consistently strong people get fewer simulations.
Training overdue
A localised, factual reminder, then compliance status and any escalation you permitted.
Entered a high-risk group
More frequent checks, always within contact caps.
dashboard.cyberfrogsecurity.com/human-risk/interventions/newTargeted interventionAutopilot handles the defaults. Step in whenever you want.Target cohortHuman Risk > 70FinanceGermanyWeaknessPayment-fraud detectionActionBEC and invoice-fraud trainingScheduleOver the next 5 working daysExclusionsSame training in the last 60 daysCohortRe-evaluated continuously38 people qualify todayPreview who, what and when before anything is sent.Respects caps, working hours and cooldowns.Approve intervention
dashboard.cyberfrogsecurity.com/human-risk/people/anna-kellerRisk Journey: Anna KellerWhy the score changed, what happened next, and whether it helped.Phishing simulation deliveredInvoice fraud, difficulty 3 of 512 OctSimulated credentials submittedVerified action, behavioural risk up12 OctMicro-lesson completedCredential theft, quiz passed12 OctFollow-up simulation scheduledNew scenario, same skill, respects cooldown26 OctReported the follow-upVerified report, behavioural risk down27 OctADMIN CONTROLSPauseRescheduleExcludeReplace activityCorrect an eventEvery change is audited and the score is recalculated.
06Risk Journey

Why it changed, what came next, and whether it helped.

Every employee has a Risk Journey: a timeline of events, interventions and outcomes. Authorised admins can pause, reschedule, exclude, replace an activity or correct a wrong event.

Audited
Every manual change is logged, and corrected scores are recalculated.
Retention checks
Short reassessments weeks after a lesson show whether learning lasted.
Effect measured
Improvement is proven by later behaviour, not by course completion alone.
See Human Risk and analytics
Security training, not surveillance

Built to be fair to employees.

Individual behavioural records need trust. Governance is part of the feature, not an afterthought.

  • Transparent

    Clear explanations for employees of what is collected, how risk is measured and what happens automatically.

  • Proportionate

    Collection limited to the purpose, with retention periods you configure.

  • Private by role

    Risk classifications only for authorised staff; reminders to employees stay factual.

  • Correctable

    Wrong events can be challenged and corrected, and scores are recalculated.

  • Human approval

    Sensitive escalations and actions affecting other systems wait for a person to approve.

  • Separated

    Each customer’s histories, rules and actions stay in its own workspace, also for MSPs.

Works with the rest of the platform

The execution layer of human risk.

One campaign that keeps getting smarter.

In a 30-minute demo we set the goal and guardrails together and walk through the first journeys.

Book a demo