Rework
Weak framing and unchecked generation accelerate the wrong direction before it is challenged.
Measure: revision cycles and discarded output02 · The executive case
Turn AI adoption into an accountable design capability.
The leadership decision is not whether designers will use AI. It is whether the organisation will define how quality, judgment, evidence, and responsibility travel through that work.
The cost of the operating gap
Unmanaged speed creates hidden work.
The largest costs often appear downstream: rework, inconsistent experiences, untraceable decisions, capability drift, and avoidable risk.
Inconsistency
Different tools and review standards fragment experience quality across teams and products.
Measure: pattern drift and quality varianceDecision risk
Unsupported assumptions become polished artefacts that are difficult for stakeholders to interrogate.
Measure: evidence coverage and overridesCapability erosion
Teams optimise prompting while core research, critique, ethics, and craft become less deliberate.
Measure: anchor-skill maturityExpected organisational outcomes
Faster resolution
Reduce time spent gathering, formatting, and reproducing information so teams spend more time deciding.
Track time to signed charter, validated concept, and resolved issue.Visible quality
Use shared dimensions, evidence standards, checkpoints, and gates instead of relying on individual preference.
Track task success, clarity, trust, recovery, inclusion, and drift.Clear accountability
Name the designer who owns each non-delegable judgment and retain the reasoning behind the decision.
Track sign-offs, corrections, overrides, and audit completeness.Durable capability
Develop skills that remain valuable as models, vendors, and interfaces change.
Track maturity by stage and anchor skill, not tool certification.Strategic differentiation
What TADF is, and what it is not.
The category is an operating system for human–AI design work, not another prompt collection or replacement narrative.
Not a tool directory
Tools are current-state implementation choices. Skills and checkpoints are the durable system.
Not an automated design process
Every stage contains work AI should accelerate and decisions it must never own.
Not a renamed Double Diamond
Nine stages reflect the new orchestration, evidence, governance, and continuous-learning requirements.
Not certification theatre
Level 3 capability is demonstrated at a live checkpoint, including knowing when not to delegate.
Enterprise risk model
Experience risk
Inaccessible, confusing, inconsistent, manipulative, or fragile experiences.
Protected by quality dimensions and validation gates.AI risk
Unsupported output, misplaced trust, hidden uncertainty, unsafe automation, and model drift.
Protected by boundaries, confidence signals, and monitoring.Organisational risk
Fragmented practice, undocumented decisions, capability gaps, and unclear accountability.
Protected by shared stages, skills, audit trails, and governance.Commercial risk
More activity without measurable value, higher rework, or AI spend without operating change.
Protected by baselines, outcome measures, pilots, and adoption reviews.What a leader is deciding
Do we want AI use to remain an individual behaviour, or become an organisational capability?
A credible adoption decision needs a named sponsor, a bounded pilot, baseline measures, two priority stages, anchor-skill assessment, governance ownership, and a review date.
Plan the adoption ↘