Eqwitty AI ReadinessOperational: the third AI maturity level
Operational organizations have AI in production and can point to value. They have also discovered how hard it is to prove that value, to keep models healthy, and to scale a win from one team to the whole business.
The diagnostic deliberately maps this level onto the Trough of Disillusionment. Many organizations reach production and then stall because momentum sags exactly when the unglamorous work of operating model and data begins. Knowing the dip is normal is half of getting through it.
What the Operational level means
Operational · Builder
Overall score 2.6 – 3.4 on the 1–5 scale
AI is live and paying off in places. Now make the wins repeatable.
On the AI curve: Trough of Disillusionment
You're in production, but proving value is hard and momentum can sag here. This dip is normal — the organizations that push through invest in operating model and data, not more pilots.
The 90-day plan at this level
Direction: run AI as a portfolio. The next 90 days industrialize the foundations — shared data, service levels, tiered governance — so scaling your winners gets cheaper, not riskier, every quarter.
Priority 1 · AI Use-Case Portfolio
Run AI as a portfolio: scale, kill, double-down.
- Quarterly portfolio review with kill criteria explicit and enforced.Measure: Review on calendar; ≥ 1 kill enacted per quarter.
- Re-baseline the dollar value of every production use case using the latest data.Measure: All production use cases have a current $ ROI.
- Pick 2 use cases to scale 5×; identify what blocks each from doing so.Measure: 2 scale plans approved.
Priority 2 · Data Foundations
Industrialize the data foundations production AI depends on.
- Stand up a feature store or shared data layer for the highest-scaling use cases.Measure: Feature store live; ≥ 2 use cases on it.
- Define the SLA for data freshness, quality, and availability for production AI.Measure: SLA written and agreed by data + product.
- Resolve the top-3 known data-quality issues blocking scale.Measure: 3 DQ issues closed.
Priority 3 · AI Governance & Security
Lift governance to the production-AI risk surface.
- Run a model-risk assessment on every production model and document mitigations.Measure: Every model has a risk record.
- Stand up a tiered approval process based on risk (low/medium/high).Measure: Process documented and applied to 3+ models.
- Run an incident response tabletop for a plausible AI failure.Measure: Tabletop completed; gaps logged.
What you will need to stand up
- A shared data layer (a ready-to-use data library for AI teams) — Stop every team rebuilding the same plumbing; let the scaling use cases share it.
- Data service levels — Agreed freshness, quality, and availability so production AI is not surprised by upstream changes.
- Tiered AI governance + risk records — Match oversight to risk so scaling never outruns your controls.
- A portfolio view of AI spend vs value — See which use cases compound and which to kill at the quarterly review.
This is the stage template. Your own plan is personalized against your answers, your weakest pillars and your industry.
Archetypes at this level
Two organizations at the same level can have very different profiles. The diagnostic names an archetype from the shape of your result.
The five levels
Band thresholds and the scoring math are on the methodology page.
Find out where you stand
A 15-minute diagnostic, then your nine-pillar score, peer benchmarks and a 90-day plan you can act on Monday. One-time payment of CA$19.80 per report.