Wave 0: the steady base that ends pilot purgatory

Wave 0 is the steady base. You secure what AI can see and do, evidence responsible use, control cost against a frozen case, reclaim shelfware, then change behaviour in cohorts. That order is deliberate.

Gartner’s 7 April 2026 I&O press release reports that only 28 percent of AI use cases fully succeed and meet ROI expectations, while 20 percent fail outright, from a survey of 782 I&O leaders in November and December 2025. Melanie Freeze, Director Research at Gartner, is clear: “High-performing I&O leaders start with realistic AI business cases and upfront preparation.” Wave 0 is that preparation in pane form.

Why sequence matters

Ambition without fit stalls

Freeze notes that failure often comes from initiatives that are overly ambitious or poorly scoped, and from expecting too much, too fast. Sequence forces fit before scale.

Usage theatre hides the gap

McKinsey’s State of AI 2025 finds about 88 percent use AI in at least one function, yet only about 39 percent report any enterprise EBIT impact. Seat counts will not close that gap.

Security first

Harden before broad Copilot

Pane 04 Security goes first because overshare and agent permissions create irreversible exposure. Labels, Conditional Access, least privilege for makers, and IR extensions belong here. See Harden before broad Copilot.

Who owns risk

Name a CISO-accountable owner for AI misuse and agent misbehaviour. Mirror that owner in the GRC register. Unowned risk is how pilots become incidents.

GRC as evidence

Register before policy lectures

Pane 05 GRC stands up the AI and agent register, DPIA triggers, literacy, and a minimum viable policy. Evidence beats slogans. Bridge honestly to NIST AI RMF and your ISO notes without claiming certification you do not hold.

Literacy is a control

People who cannot spot unsafe prompts will create shadow risk. Literacy sits with GRC and Adoption together.

Cost and value control

Freeze the case

Pane 06 Cost and Value forecasts the full stack and reconciles realised value to a frozen case. Read Potential is not realised value.

Steward challenge

Steward mode in Counsel challenges Pollyanna ROI. If you cannot name benefit owners, you are not ready to scale spend.

Licence and adoption

Use it or lose it

Pane 07 Licence runs 30/60/90 reclaim cycles. Licensed is not active. Active is not valuable. Doctrine lives in Use it or lose it.

Habit over headcount

Pane 08 Adoption uses waves, champions, and clinics. Measure changed work, not seat volume. Pathfinder ideas and deep task redesign stay fenced until Security, GRC, and cost controls hold. A thin SEE pass may still run in parallel.

FAQ

Is Wave 0 only for Microsoft Copilot?

No. The sequence fits any seat-and-agent estate. Copilot is a common trigger, not the limit.

How long does Wave 0 take?

Foundation packages often run 4–8 weeks. Operate and Scale stretch the operating rhythm. See offerings for week-1 outputs.

Can SEE delay Wave 0?

It should not. Run thin awareness beside Steady work.

What does week 1 produce?

Named owners, baselines, and a prioritised action list. That is how you leave pilot purgatory early.

Lead with risk ownership, frozen value case, and reclaim before seats. Cite McKinsey use-versus-EBIT and Gartner ROI success rates as external context, not as your KPI targets.

Where next?

Start on /start, or open a Counsel session example. DIGITAL SKILL WORKS LTD / 12 PANES