Map where your team really stands with AI. Then move.
Everyone has been handed a new shape overnight, and nobody drew it. CapX is a map: a short, evidence-scored conversation with Ana that shows where each person actually stands — and the gap between what they’d say and what they can show. Multi-Modal Design is the method that moves them. The map shows where you are; the method moves you; your next reading shows how far.
Your map lands with you first. Nothing feeds a ranking, a review, or a hiring screen.
The market that judges capability admits it can’t measure it. Map it yourself instead. (Talogy, 2026)
Coverage is nearly universal. The gap barely moves. Coverage isn’t capability. (DataCamp 2026)
When the AI is switched off, the work has to still be there. That’s the gap the map draws. (Liu et al. 2026, RCT N=1,222)
Faster typing is not a transformation.
Two years of AI adoption has produced a lot of motion and very little method. Teams prompt, summarise and generate at speed — and call it capability. The measurement says otherwise from two directions at once: most companies now train their people on AI, and most still report a skills gap. Coverage without capability. And when capability is measured properly rather than self-reported, the gap is wider than anyone admits.
Underneath the noise is a plainer fear. AI hands people a new shape — the specialist told to run a team’s work, the manager told to lead twice the reports and keep shipping — and then measures nobody against it. The dread isn’t about the AI. It’s about being measured against a shape no one has drawn. So we draw it. Not to judge you against it: to let you see it, and walk toward it.
What the current tools measure, and what they miss
| What it measures | What it misses |
|---|---|
| Self-report surveys | Blind to the say-do gap by design — they only collect the “say” |
| Prompt-engineering courses | Teach the keystroke, not the lifecycle |
| Big-4 maturity assessments | A snapshot you can’t re-take; org-level, not person-level |
| LMS completion data / AI adoption dashboards | Measure activity or usage, not demonstrated capability |
| AI-native assessment agents (hiring & workforce platforms) | Score you for someone else’s decision — a verdict you may never see, not a map you can act on |
A map, not a verdict.
CapX, the capability map from Multi-Modal Design, reads where each person actually stands — on evidence, not self-belief — and hands the reading to them first.
Each person has a short, adaptive conversation with Ana. She asks for concrete things you’ve actually done — not agree/disagree statements — and follows the evidence where it goes. Short, async, on your own link.
Every answer is scored for level (1–7, aligned to SFIA 9 and DQ / IEEE 3527.1), how confident the reading is, and how concrete the evidence was. The legend under each reading is the evidence sentence — so you can see why, and appeal it if you disagree.
Before the evidence, you say where you think you are. The map shows both — and the band between them. Over-claimed, under-claimed, or aligned. The say-do gap is the one finding a survey can’t produce, because a survey only collects the “say.”
Your map comes to you first. Team read-outs aggregate from there. Nothing feeds a ranking, a performance review or a hiring screen. The reading exists so you can act on it, not defend yourself with it.
The map is not the territory. A reading is taken at a point in time; it doesn’t predict where you’ll arrive. That’s why it comes with a method — and a next move.
The instruments are on the desk. A person runs the session.
Multi-Modal Design is a way of working with AI agents across strategy, design, build, content and education — as instruments, not typists. Six stages; every one of them gated by a human who owns the decision and leaves a record.
Somebody has to run the session — decide what to hand to an instrument, where the gate sits, whether the output is right. We call that person the Producer. It isn’t a job title; in AI-augmented work everyone produces something. CapX reads the Producer stance too, so a lead can see their own map before they ask their team to.
Every fork is reasoned, logged, and auditable.
Inside the bar. Do it.
Needs something new but absorbable. Proceed, and write the decision down.
Reframes, forks, anything irreversible or outward-facing. Stop, and bring it to a named human with the context and authority to decide.
The map shows where you are. The method moves you. Your next reading shows how far.
Not a report on a shelf. A loop.
One reading tells you where you stand. Two tell you whether you moved. That delta — per person, per capability, on evidence — is the number a CFO asks for and a self-report can’t give.
- 01Assessa conversation with Ana; behavioural evidence, async.
- 02Diagnosethe gap: self-rated against demonstrated; where the fog still is.
- 03Upliftyour next move: a route per person’s shape, drawn from the method — which stage, which gate, which instrument.
- 04Re-assessthe next reading. Movement shown as counted rungs, not opinion.
Three questions, one loop.
Map the bench on evidence, staff the gates on it, run the method on Monday.
Bring your framework — SFIA, DQ, your own. The framework is data; the engine that reads evidence against it is the same.
Check your signal before you move — your first reading is free. Your map is yours. Usually you’re closer than you thought; where you’re not, there’s a route.
- a family digital-wellbeing platform — full strategy → design → build → content cycle, a five-month solo AI-augmented build
- Insyteful — the assessment-to-synthesis pipeline that became the engine
- Greenstone Group — the method in a non-product strategy context
- a cost-leadership coaching engagement — the method transferring to a non-technical team
- a subcontracted product build — scope held inside an 11-day cap, 34 decision records, independent QA approved
Aligned to the standards your organisation already assesses against.
CapX scores against public frameworks so the map reads in a language procurement, HR and government already speak. The framework is data; the reading is the same engine.
| Framework | How CapX uses it |
|---|---|
| SFIA 9 | Scored against. |
| DQ / IEEE 3527.1 | Scored against. |
| Alan Turing Institute AI Skills for Business v3 | Directional convergence, cited, not scored against. |
| OECD & European Commission AILit | Directional convergence, cited, not scored against. |
| US Department of Labor AI Literacy Framework | Directional convergence, cited, not scored against. |
Frequently asked
Founding Cohort — two ways in.
your team mapped, the method installed with JD in the room, the delta measured at 90 days. Scope capped in days, not promises.
the signal check, the open playbook, and the second reading when it ships — you’re the proof-ground and you get the map.