01 · Volume
3.4×
growth in fraud alerts per analyst over [PERIOD] [STAT]
Transaction monitoring, public reports and inter-agency referrals grow faster than headcount. Queues become backlogs; backlogs become losses.
Certification for AI-assisted security work
Powerful AI. Accountable humans.
DETECT. ANALYZE. PREVENT.
DW-GPT certifies the officers, analysts and responders who deploy, supervise and audit AI agents — so every alert, lead and decision stays explainable, lawful and owned by a human.
- 4 specialist tracks
- 3 credential levels
- Responsible AI in every path
Agent Operations
Live · Simulated
- Alerts triaged
- 12,408
- Anomalies flagged
- 1,294
- Cases correlated
- 386
- Awaiting human review
- 7
- T+02:03:20EWS-SIGNAL 0.39→ MONITOR
- T+02:02:43PHISH-CLSTR 0.74→ HUMAN REVIEW
- T+02:02:06KYC-DRIFT 0.66→ ANALYST QUEUE
- T+02:01:29TXN-BURST 0.82→ HUMAN REVIEW
Consequential actions require a named human reviewer. *Low-risk duplicates only; every merge is logged.
Built with practitioners across policing, finance and public safetyPartner logos: placeholder
- [Partner Police Service]
- [National CERT]
- [Partner Bank]
- [Financial Intelligence Unit]
- [Emergency Management Agency]
- [Fintech Partner]
- [Public Safety Institute]
- [Regional CSIRT]
01The problem → the shift
Threats move at machine speed. Most teams still triage by hand.
Volume, velocity and noise are outpacing every investigative team we speak to. Hiring alone will not close the gap — but handing decisions to unsupervised software is not an option either.
02 · Velocity
48 min
median time from initial access to lateral movement [STAT]
Attackers automate reconnaissance and exploitation. Triage measured in hours cannot contain an intrusion measured in minutes.
03 · Overload
70%
of alerts closed as false positives [STAT]
Skilled investigators spend their day clearing noise — while the signal that matters waits in the same queue.
PlaceholderFigures above are illustrative layout values, not research findings. Replace each with a sourced statistic and citation before publication.
→ The shift
Supervised agents change the equation — humans keep the verdict.
- 1
Agents do the sifting
Collection, deduplication and first-pass correlation run continuously, at machine speed, with every source cited.
- 2
Humans do the deciding
Every consequential action stops at a named reviewer with the authority to approve, amend or refuse it.
- 3
Everything leaves a trail
Prompts, data sources, model versions and decisions are logged so any outcome can be reconstructed and challenged.
Where an analyst’s time goes
Illustrative workflow — not measured data.
Without agents
Collect · 34%
Triage · 30%
Correlate · 26%
Decide · 10%
With supervised agents
Review evidence · 30%
Challenge & verify · 30%
Decide · 40%
Human decision time Human review of agent work Manual processing
02Certification tracks
Four specialisms. One standard of accountability.
Each track pairs hands-on agent tooling with the judgement to supervise it. Every track includes the mandatory Responsible AI core.
Select a track to see modules, skills, prerequisites and exam format.
- TRK-01
AI-Agent Policing & Investigations
Put investigative agents to work on case triage, evidence correlation and OSINT — without breaking chain of custody or disclosure duties.
- Agent-assisted case triage
- Evidence correlation
- OSINT workflows
- + 2 more modules
6 weeks · ~48 h - TRK-02
Financial Fraud Intelligence
Deploy transaction-monitoring and AML/KYC agents that surface real risk, map mule networks and keep investigators firmly in the loop.
- Transaction-monitoring agents
- AML/KYC automation
- Anomaly detection in practice
- + 2 more modules
6 weeks · ~52 h - TRK-03
Cybercrime Response
Orchestrate agent-driven triage, incident response and forensic support — with approval gates on every containment action.
- Agent-driven threat triage
- Incident response orchestration
- Digital forensics support
- + 2 more modules
8 weeks · ~64 h - TRK-04
Early Warning Systems
Fuse multi-source signals into calibrated early warnings, prioritise alerts sensibly and escalate crises through human-owned playbooks.
- Predictive risk signals
- Multi-source data fusion
- Alert prioritisation
- + 2 more modules
6 weeks · ~46 h
→ What you earn
Every track ends in a signed, verifiable credential.
Sample certificates with fictional identities. Open one to view it full screen, switch tracks, or print it.
03Certification levels
From operator to auditor. Earned, not attended.
Three levels, each assessed in the sandbox and in front of people. Credentials expire so that skills keep pace with law, policy and tooling.
- L1 · Level 1
Associate
Operate supervised agents safely within established procedures.
- Hours
- 40
- Labs
- 8
- Valid
- 2 years
Requirements
- Responsible AI core module
- Foundation units of one track
- Pass the Associate exam (scenario lab + written)
Renewal: continuing professional development + update assessment.
- L2 · Level 2
Professional
Design and lead agent-assisted workflows for a team.
- Hours
- 90
- Labs
- 18
- Valid
- 3 years
Requirements
- Associate credential
- Complete one full certification track
- Scenario lab, written exam and oral defence
Renewal: continuing professional development + update assessment.
- L3 · Level 3Highest
Expert
Govern, audit and assure AI-agent programmes across an organisation.
- Hours
- 160
- Labs
- 30+
- Valid
- 3 years
Requirements
- Professional credential in two tracks
- Capstone: audit a live-like agent deployment
- Peer board review and ethics interview
Renewal: continuing professional development + update assessment.
04Interactive lab preview
Watch an agent reason. Then you decide.
This is how every DW-GPT lab ends: the agent gathers, correlates and scores, then stops. Nothing consequential happens until an accountable human reviews the evidence. Your call is graded against an assessor's, and not every agent is right.
2 · Run the agent
Simulation · All accounts, people, places and events are fictional and generated for training. No real data is used or transmitted.
agent-console://sim/SIM-FIN-0192
Simulation$ agent --case SIM-FIN-0192 --mode supervised
Ready. Press “Run simulation” to begin.
--
Confidence
—
Evidence surfaced · 0/4
05Mandatory core module
Responsible AI in Law Enforcement
Capability without accountability is a liability. This module is required in every track and assessed at every level — because the people affected by these systems deserve decisions that are lawful, fair and explainable.
“The agent can find the pattern. Only a person can be answerable for what happens next.”
- 01
Human oversight
Agents recommend; named, accountable people decide. Every consequential action passes a human checkpoint with the authority to stop it.
- 02
Bias & fairness
Test data and outputs for disparate impact before deployment and continuously after. Know which communities your error rates fall on.
- 03
Privacy & data protection
Collect the minimum, retain only as long as lawful, and apply purpose limitation to every dataset an agent can reach.
- 04
Legal admissibility
Understand disclosure duties, expert-evidence rules and how AI-assisted findings are tested and challenged in court.
- 05
Explainability
If you cannot explain why an agent flagged something, you cannot act on it. Require reasons, sources and confidence with every output.
- 06
Audit trails
Log prompts, sources, model versions, reviewers and decisions so any outcome can be reconstructed, scrutinised and corrected.
06Learning experience
Practice where mistakes are free. Prove it where it counts.
Built for working professionals: flexible self-paced labs, live cohort sessions, and credentials that stand up to scrutiny.
Sandbox labs
Isolated environments with synthetic case data, realistic tooling and resettable state. Break things safely; nothing touches production or real people.
Scenario simulations
Branching exercises where the agent is sometimes wrong — on purpose. You learn when to trust, when to verify and when to refuse.
Instructor-led cohorts
Small cohorts led by practitioners from policing, financial crime and incident response, with live debriefs.
Digital badges
Shareable badges for each level and track, with the skills evidenced listed in the badge metadata.
Verifiable credentials
Cryptographically signed certificates that employers, courts and auditors can verify in seconds.
Progress analytics
Individual and team dashboards: lab performance, oversight decisions and readiness by competency — ready for training directors and procurement reviews.
07Credential verification
Verifiable in seconds. By anyone who needs to know.
Every DW-GPT credential is signed and publicly verifiable, so employers, courts and auditors can confirm competence without a phone call.
Awaiting certificate ID
08Voices from the field
Trusted by people who have to defend their decisions.
- Placeholder testimonial
The oral defence was the hardest part — and the most useful. My team now asks “why did the agent flag this?” before anything else.
[Name Placeholder]Detective Inspector, [Partner Police Service]
09Pricing & enrollment
Seats for individuals, teams and whole agencies.
Transparent per-seat pricing for teams; procurement-friendly terms for government and enterprise.
IndicativePrices in USD, excl. tax. Final terms confirmed at enrollment.
Individual
For professionals certifying on their own.
$39/ month
Billed annually
- One certification track of your choice
- Responsible AI core module
- Sandbox labs and scenario simulations
- One exam attempt per level
- Verifiable credential and digital badge
- Recommended for teams
Agency / Team
Train units and departments together.
$31/ seat / month
$775 per month for 25 seats
550 · −10%100 · −15%250- Everything in Individual
- Access to all four tracks
- Scheduled cohorts for your team
- Team progress and readiness analytics
- Seat management and admin roles
- Invoicing and purchase-order billing
Enterprise / Government
Agency-wide programmes with sovereign controls.
Custom
Framework and tender-ready quotes
- Everything in Agency / Team
- Sovereign or on-premises sandbox options
- Curriculum aligned to your SOPs and legal framework
- Closed, instructor-led cohorts
- SSO, audit exports and data-residency controls
- Procurement support: [FRAMEWORK AGREEMENT PLACEHOLDER]
10FAQ
Questions procurement will ask.
Can’t find what you need? Book an agency demo and we’ll walk your team through the programme.
No. The program is built for practitioners, not engineers. You will configure, supervise and audit agents through purpose-built interfaces. The Cybercrime Response track assumes working knowledge of logs and networking; the others assume only your professional experience.
READY WHEN YOUR TEAM IS.
Put AI to work. Keep humans in command.
Enroll individually, or talk to us about certifying a unit, a department or an entire agency.
The monthly briefing
Cohort dates, curriculum updates and policy changes affecting AI in enforcement. No spam; unsubscribe anytime.