ai · security · skills

How it works

Build the capability one team at a time.Prove it before you scale.

Pick one security team. Fit a little AI help to the way it already works, and measure what changes. No big-bang, and nothing to rip out.

One team at a timebaseline
Pick one security function
Run the function diagnostic
Save the baseline: maturity × autonomy, the gate reading

Assess: One function, honestly measured. The diagnostic sets the baseline everything after is judged against.

The idea

A way of working, not a tool to buy.

Every team is picking up AI on its own. You can’t secure what you can’t see, and you can’t fund what you can’t measure. So instead of one big programme, this builds real capability the safe way: one team, one job, one measured result at a time. One questionnaire runs both tracks: it grades the team, and it names what each person on it learns next.

The method

Three short steps.Each one ends in a go or no-go.

You only fund the next step once the last one has paid off. That’s what keeps it low-risk.

Phase 1 · A few weeks

Assess

Baseline one team on evidence — where it really stands today. Nothing changes in your tools.

You walk away with

A one-page read: the top gaps and the one or two moves worth funding.

GO?Pays off? Fund the pilot.

Phase 2 · A few weeks

Pilot

Fit one helper to that team’s real work, and capture the number before and after.

You walk away with

A measured before-and-after, and a helper you keep and can reuse.

GO?Proven? Fund the scale-up.

Phase 3 · Ongoing

Scale

Roll the proven pattern across the team, then on to the next one.

You walk away with

A repeatable pattern — the next team is faster than the first.

Why assess · the loop

You assess to earn the next rung of autonomy.

The diagnostic reads a function on two axes: how well its AI is governed, on the AI Security Maturity Model (AISMM, L1–L5), and how far its work is automated, on the AI Cyber Maturity Model (AI-CMM, L1 Manual→L4 Autonomous; our model · calibrated to SAE J3016, never a CSA or SAE rating). The gate connects them: autonomy must never outrun governance, because autonomy without measured governance is unpriced risk.

The industry frames it the same way. Forrester’s AEGIS (Agentic AI Enterprise Guardrails For Information Security) framework argues for least agency, granting an agent only the autonomy your controls can catch: that is our gate. It argues for continuous assurance, re-verifying as systems change: that is our loop. External framing, cited not copied; the rubric underneath stays CSA (Cloud Security Alliance) AISMM and the AI Controls Matrix (AICM).

1 · Assess2 · Fix the top gaps3 · Re-assess4 · Advance one rungone rung per loop

Cadence: re-assess when something material changes (a new model, a new data class, a rung-advance attempt), not on a calendar.

What each rung advance requires

ManualAssisted

Gate clear: governance at L2 (Repeatable) for the workflow being assisted.

A human still approves every action, so written, repeatable procedure is the floor.

AssistedAugmented

Gate clear, plus Security Monitoring ≥ L3 and Incident Response ≥ L2.

Running AI in the loop demands monitoring and incident response in place — you must see what it does and respond when it errs.

AugmentedAutonomous

Gate clear, plus Security Monitoring ≥ L4 and Incident Response ≥ L3 and Model Security ≥ L2.

Letting AI act autonomously additionally demands model-security controls — adversarial testing and artifact integrity — before per-action human approval is removed.

Each pass hands youThe maturity radarA ranked gaps registerA fitted-few prescriptionA saved baseline you re-run

For leaders

You can’t fund what you can’t measure.

Every step is graded on evidence, costed, and proven before the next one is funded.

A starting point you can defend

Where the team stands today, graded on evidence — not opinion. That’s the number you’re improving from.

A next step with a price tag

The one or two moves worth funding, each tied to a specific safeguard. You know exactly what you’re buying.

A pattern you can repeat

What works on one team becomes the recipe for the next — so the second team is faster than the first.

The one rule

How independently AI is allowed to act never gets ahead of how well it’s governed. We call it the gate — and it’s the reason the climb from “AI drafts, a human approves” to “AI acts, humans audit” stays safe.

See it work

Two jobs run end to end: a Security Operations team clearing its queue, and an AI app’s guard against jailbreaks. New to all this? Start with the AI threats in plain English →

Anchored to

Recognised standards — the CSA control library and maturity model — mapped to the rules you answer to (the EU AI Act, ISO/IEC 42001, NIST), so satisfying one clears many. See the crosswalk →

On the numbers. Every figure is estimated until it’s measured on your own data, on a staging copy, with no production or real personal data involved.

Start with one team.

Pick a function, run the short diagnostic, and see where it stands — in a couple of minutes.