The AI & Data Lens
Pilots Demonstrate Capability. Designed Processes Produce Results.
The constraint on enterprise AI is rarely the model. It is that the work it is meant to improve was never defined and the data it is meant to read was never governed.
Steve Kopecky · 5 minute read
Why capable pilots change nothing
A pilot proves that a capability exists. It says nothing about whether the enterprise can absorb it. Absorption requires a defined process to insert the capability into, a measure that will register the improvement, and an owner accountable for the outcome rather than the technology.
Where those are missing, the pilot produces an impressive demonstration and a quiet return to the previous way of working. The organization concludes that the technology is immature. Usually the technology was the only mature element in the exercise.
AI applied to an undesigned process automates a decision nobody can explain.
Governed data is a prerequisite, not a workstream
Models inherit the discipline of the records they read. Where the customer master has three versions of the same account and the item master carries obsolete records nobody may delete, the output is confidently wrong — and confidently wrong is considerably more expensive than obviously wrong.
Governance here means something concrete: a named owner per data object, a rule for creation, a rule for retirement, and a measure of record quality that someone reviews. Four things, rarely all present.
Name the measure before the use case
The most useful discipline in an AI portfolio is to require, for every candidate use case, the measure it is expected to move, its current value, and who owns it. Candidates that cannot supply all three are research, not initiatives, and should be funded and governed as research.
This single filter typically removes half of an AI backlog and doubles the credibility of what remains — which is the outcome a board is looking for when it asks what the investment has produced.
Exhibit
Exhibit — The AI candidate filter
One row per candidate use case. A blank in any of the first four columns means the candidate is not ready to fund as an initiative.
| Use case | Process it sits in | Measure it moves | Current value | Data owner | Decision |
|---|---|---|---|---|---|
| Quote drafting assistance | Order intake — defined | Quote turnaround | 3.5 days | Commercial | Fund |
| Demand forecasting | Planning — partially defined | Forecast error | Not measured | — | Hold |
| Service triage | Support — defined | First-contact resolution | 61% | Service | Fund |
| Document summarisation | No named process | None stated | — | — | Research |
The filter is not a governance formality. It is the difference between an AI portfolio and a collection of demonstrations.
Before the next pilot is funded
- The process the capability will sit inside is defined as performed.
- The measure it is expected to move is already being measured.
- Each data object it reads has a named owner and a retirement rule.
- An accountable business owner exists, distinct from the technology sponsor.
- The organization has capacity to adopt the change, not only to trial it.
AI rewards enterprises that have already done the unglamorous work. That is inconvenient, and it is the entire finding.
Share this
Share on LinkedInAligned to FOUNDATION™
Who are we, and what business are we actually trying to build?
Present reality, leadership capability, purpose and organizational health.
Article
The Architecture That Created Today's Success Rarely Produces Tomorrow's Results
The structure that carried a business to its current size is usually the constraint on its next stage. That is a design problem, not a people problem.
6 minute read
Article
What Is Enterprise Business Architecture — and Why Does It Determine Performance?
Enterprise business architecture is the deliberate design of how an organization decides, aligns, and executes. It sets the ceiling on performance long before effort does.
9 minute read
Assessment
FOUNDATION™ Diagnostic
A structured read on leadership capability, clarity of purpose, trust and organizational health.
Maturity read on screen, a print sheet, and an emailed copy.
Assessment
Lifecycle Stage Diagnostic
Fifteen statements that locate the business on the curve — Honeymoon through Reinvention — and name the binding constraint.
Stage placement, constraint, and the first move for that stage.
Instrument
30 / 60 / 90 Day Action Plan
Turns the lifecycle result into sequenced work: the stage supplies the architectural move, the lowest-scoring domain supplies what removes the constraint. Owners, dates and check-off included.
A checkable plan on screen and a board-quality PDF of the three horizons.
Instrument
Lifecycle Before & After Sheet
Holds the diagnostic result as the before, the stage and domain targets being architected as the after, the operating measures the change is judged against, and the reinvention actions carried from the ninety-day plan.
A one-page board-quality before-and-after PDF, plus an emailed copy.
Podcast
Kristina George — How an office manager became an owner leading a team of eighteen.
Kristina George describes the move from operator to owner and the decisions that made it hold. The conversation examines anchoring the agency on explicit values, investing in employee engagement, and simplifying the operating model so growth did not dilute the culture.
Podcast
Mike Domitrz — How a mission born from tragedy became a leadership standard.
Mike Domitrz shares the origin of The Center for Respect after his sister's assault in college, and how the work grew into a program trained across universities, high schools, the military, and businesses. The conversation examines how respect becomes an observable leadership behavior rather than a value on a wall.
Guide
Read the Curve Before You Build the Next One
The second curve is a timing decision before it is a strategy decision. Most companies read the curve years late, at the point where the options are cheapest to see and most expensive to take.
7 minute read
Engagement
Compass Executive Forums
A confidential peer table of owners and executives working the same architecture, month after month.
Monthly, by invitation
Engagement
The Compass Operating System™
The five-discipline architecture Compass designs and installs with a leadership team.
The authoritative source
Continue reading
The Architecture That Created Today's Success Rarely Produces Tomorrow's Results
The structure that carried a business to its current size is usually the constraint on its next stage. That is a design problem, not a people problem.
Growth & Value CreationThe Second Curve Begins Before the First One Ends
The right moment to build what comes next is while the current business is still strong — which is precisely when no one feels the need to.
SimplificationComplexity Quietly Consumes Margin, Capacity and Leadership Attention
No one approves complexity. It accumulates one reasonable decision at a time, and it is paid for in margin, speed, and executive attention.
SimplificationFOCUS: The Vital Few Constraints That Create Economic Value
FOCUS is not doing less. It is concentrating effort on the customers, products, and markets that produce disproportionate value — and removing the complexity mirror they cast inside the organization.
Organizational ArchitectureWhat Is Enterprise Business Architecture — and Why Does It Determine Performance?
Enterprise business architecture is the deliberate design of how an organization decides, aligns, and executes. It sets the ceiling on performance long before effort does.
Organizational ArchitectureThe Ten Elements of an Enterprise Architecture — and Why They Only Hold Together
Strategic intent, simplification, leadership, governance, planning, cadence, decision rights, performance, enabling systems, and renewal. Designed together, or not designed at all.
Strategy & ExecutionArchitecture vs. Improvement: Why Better Execution Cannot Fix a Flawed Design
Improvement makes the current system perform closer to its limit. Architecture changes the limit. Confusing the two is the most expensive mistake in the mid-market.
Organizational ArchitectureEvery Organization Is Producing Exactly What Its Design Allows
Decision rights, cadence, reporting lines, incentives and leadership habits are not background. Together they are the machine producing the number on the books.
Organizational ArchitectureEffort Cannot Solve a Design Problem
Pressure produces a short improvement, then the system pulls the result back to where the architecture says it belongs.
Organizational ArchitectureArchitecture Is Designed, Not Discovered
The work is concrete: name the few outcomes that matter, place each decision where it is properly made, remove the complexity consuming margin, and rebuild the cadence so commitments close.
The CEO / Owner LensThe Chief Executive Is Not Supposed to Be the Integration Mechanism
When strategy only reaches the work through the chief executive, growth adds volume to an unresolved design. The fix is structural, not personal.
The CHRO / People LensTurnover Is Usually a Structural Reading, Not a Hiring Result
When roles, capability, managers and recognition point in different directions, capable people leave — and the organization reads it as a recruiting problem.
The Operations LensPerformance That Depends on Who Is on Shift Is Not Performance
Where the process is undefined, capable people carry it. The result looks like variable performance and is read as a discipline problem.
The ERP / Systems LensAn ERP Program Is a Process Decision Wearing a Technology Budget
Systems enable the management system; they never replace it. Configuring before the process is designed automates the current mess at scale.
The Exit / Succession LensA Buyer Discounts Precisely What the Owner Cannot Hand Over
Transferable value is documented processes, governed data, named successors and results that hold when the founder is not in the room.
The Leadership Team LensA Team That Meets Often and Decides Rarely Is Compensating for a System
When reviews end in discussion rather than dated commitments, the same root cause is rediscovered every quarter.
The Transformation LensExhaustion Is Routinely Misread as Resistance
Change fails at the point of absorption far more often than at the point of design. Capacity, honesty about recent change, and reinforcement decide the outcome.
The Strategic Program LensBecoming Efficient at Work You Never Chose
Portfolios fail at intake far more often than in delivery. Unranked work consumes the capacity that strategy was supposed to direct.
The Next-Curve LensThe Next Curve Is Absorbed by the People Already at Capacity
A next curve is designed on paper and delivered by an organization that must also run the current one. Readiness is whether the system can carry both.