What training platforms now automate with AI, and what to check

Almost every training platform on this site now describes some use of AI. The uses differ a lot, and so do the questions worth asking about them.

Where does AI appear?

In building content. Immersive Labs announced AI Lab Builder 2.0 in January 2026, alongside an AI-Powered Crisis Simulator for custom scenarios. Conducttr describes LISA, which it calls a text-to-MEL chatbot, for drafting an exercise's timeline of injects. TryHackMe describes tabletop exercises generated from your industry, attack vector and tech stack, which teams can edit or re-prompt.

In playing roles. Reflex Security describes an unscripted adversary that reacts to each decision, AI agents that play threat actors, journalists, board members, legal counsel and regulators or fill absent seats including the CEO's, and an AI facilitator, Victoria, that joins Zoom, Meet or Teams. Immersive Labs describes AI actors in its crisis simulations that players can question.

In helping learners. Immersive Labs announced an AI In-Lab Assistant in January 2026.

In being tested. Some ranges now test AI agents as well as people: Cloud Range launched an AI Validation Range on 24 September 2026, and SimSpace invites teams to bring their AI agents into the environments it builds.

In measuring. iluminr's August 2026 update added an AI Context Engine that uses business continuity and GRC data during scenarios, and behavioural evidence such as decision velocity, confidence and alignment.

Why does it matter for the choice of training type?

Mostly because of preparation time. A traditional tabletop takes weeks to write, which is one reason many organisations run one a year. Reflex compares a traditional tabletop's 2–4 weeks of preparation with minutes on its platform; that is the vendor's own comparison, but the direction is the point. When scenarios are generated, exercises can run more often, and frequency is what turns findings into fixes.

What is AI doing to the skills being trained?

Hack The Box's benchmark research gives a view. Its 2026 Global Cyber Skills Benchmark brief, published on 26 August 2026, says 68% of the top 25 teams used AI agents and that median solve time fell from 26.1 hours in 2024 to 13.8 hours in 2026. An earlier Hack The Box report, covered in March 2026, found AI-using teams reached 27% challenge success against 16% for the top human-only teams, and warned of a talent-pipeline risk in mid-tier tasks. Training for analysts increasingly needs to include working with AI tools, not only around them.

What should you check?

1. What does the AI decide, and what does a human review? A generated scenario should be checked by someone who knows your environment.
2. What data does it use? Scenarios built from your environment are more realistic, and you should know which data leaves your hands. Reflex describes using public footprint information and optional telemetry; iluminr describes using business continuity and GRC data.
3. Can you switch it off? Immersive Labs describes an AI opt-out setting.
4. How are AI-played roles shown in the report? An after-action report should make clear which decisions came from people.

What next?

Go back to lesson 1 if you have not mapped your gaps yet, or read the Range Desk notes for dated coverage of what vendors announced in 2026.