Band-Aids, Six Sigma,
and Fire Codes
One bad habit. One proven fix. One analogy your aunt will remember.
Molly Woodbridge, Systems Producer, Fort Worth
Arrow keys, space, or the buttons below move through the deck. Some slides build one line at a time.
Public-trust institutions buy AI.
Institutionalizing it is different work.
The gap is between those two sentences.
The standards exist. NIST wrote the framework. Texas wrote the rules and assigned every covered agency and local government an AI Risk Officer. The software exists. Every vendor will sell you a seat tomorrow.
The gap sits between them: the staffed, funded, repeatable work of turning assigned duties into reviewed decisions, kept records, and workflows that survive turnover.
The gap between policy and practice becomes the Ungoverned Middle. Each organization is left to staff it, fund it, and figure it out.
NIST AI RMF 1.0 · Texas SB 1964, effective Sept. 1, 2025 · 1 TAC 219.21(a), effective Mar. 18, 2026
The Band-Aid Cycle
Skip the upstream baseline. Pay the downstream tax. Budget the tax as if it were normal. Repeat.
Skip the upstream baseline
It looks like savings. It is a loan.
Problems compound quietly
Interest accrues where nobody is looking.
Fund the downstream fixes
Now with staff, programs, and urgency.
Call the tax "the budget"
The loan payment becomes a line item.
Repeat until the budget is entirely tax. It is not a failure of caring. It is a failure of sequence.
Press the arrow to build the four steps
Education
The skipped upstream
- Proper nutrition
- Physical movement and unstructured play
- Allowing kids to regulate their bodies during the school day
- Early health care
The downstream tax
- Remedial tutoring
- Behavioral intervention programs
- High school retention staff
- Administrative overhead that outgrows instruction
The upstream side is not hypothetical. There is more than a decade of peer-reviewed research on movement and play as baseline conditions for learning.
We fund the fix at full price and call the prevention unaffordable.
Workforce
The skipped upstream
- Straightforward job information
- Realistic fit screening
- Non-negotiables named before applying
The downstream tax
- Churn and ghosting
- Mis-hires and burnout
- Re-recruiting the same seat
Job transparency is nutrition for the labor market. Skipping it has a tuition rate.
A workforce application. GateForce85 scores a job posting against a person's real-life constraints and answers ADOPT, PARK, or REJECT with the reasons. Audits postings, never people. Working browser prototype. Read more
AI Adoption
The skipped upstream
- Who reviews the outputs
- Who owns the decisions
- What gets documented, and how you leave
The downstream tax
- Cleanup and do-overs
- Dependence on one vendor's defaults
- Institutional memory that evaporates
The tools arrive, the training happens, and everyone assumes implementation has occurred.
An organization can own software and still not own an operating layer.
One pattern. Three invoices.
Education
Skip nutrition, movement, early care.
Pay for tutoring, intervention, retention.
Workforce
Skip straightforward information and fit.
Pay for churn, burnout, rehiring.
AI Adoption
Skip the operating discipline.
Pay for do-overs, lock-in, lost memory.
Three sectors, one recurring sequence: skip the conditions that make the work possible, then fund the consequences.
Press the arrow to build the three sectors
Has anyone ever fixed
a problem shaped like this?
Twice. Once with a Greek letter, once with a clipboard.
Six Sigma: a discipline, not a product
1986. Motorola. An engineer named Bill Smith connects field failures to factory defects and proposes something unglamorous: a shared way of measuring, naming, and removing defects before they are sent. Software did not make it work. A method was institutionalized. The tools came and went. The discipline stayed.
defects per million opportunities: Six Sigma's conventional benchmark.
The habit it replaced: inspect at the end, fix in the field, call it quality.
How a discipline actually scales
Practitioner
Bill Smith, a working engineer. Not a committee.
Real conditions
Motorola's actual defect data. Not a whiteboard.
An executive champion
CEO Bob Galvin made it the company's language.
Public validation
The Malcolm Baldrige National Quality Award, 1988.
Then, scale
GE, Honeywell, and every conference budget for a decade.
A sequence, funded: more than 45 million dollars a year in training, 40 percent of it for quality.
Motorola Annual Report, 1988
Press the arrow to build the five steps
What Six Sigma built as it scaled
Coined vocabulary
Sigma levels, then belts and DMAIC. Language that traveled without its author.
Portable documentation
Methods written down, so the discipline survived the founder.
Repeatable frameworks
The same process worked on the next problem. And the next.
Six Sigma assembled these over a decade. Human Capacity Infrastructure starts with all three: working language, documented frameworks, repeatable structure. It is the checklist. The rest has to be earned.
Press the arrow to build the three
Six Sigma showed how a discipline scales. It does not explain Human Capacity Infrastructure at Thanksgiving.
For that, we need fire codes.
Nobody calls a fire code anti-building
Fire codes did not appear because nobody had buildings. They appeared because everyone had buildings, and each one burned in its own creative way.
Whole cities burned before shared standards existed. The response was not to ban buildings or trust each builder's personal enthusiasm for safety. It was a common standard that made every building legible, inspectable, and survivable.
The lesson survived because it was boring, written down, and enforced. Aspirations were not involved.
A code:
- Puts a name on responsibility
- Sets limits before occupancy, not after the fire
- Inspects on a schedule
- Records what failed and what was fixed
- Outlives the people who wrote it
In practice the duties spread across owners, operators, and inspectors. That distribution is the system, no single hero.
Human Capacity Infrastructure proposes the same jobs for AI adoption.
Press the arrow to build the list
Human Capacity Infrastructure is fire codes for AI adoption
The building is the tool. The code is whether it holds.
Press the arrow to build the rows
AI can be affordable to try and expensive to govern.
Prices verified October 2026 from each vendor's own pricing page; they change. * Lowest published enterprise entry: Claude Enterprise, $20 per seat per month billed annually, 20-seat minimum, usage billed separately; most enterprise tiers publish no price. Standard seats $20 to $25, premium $100 to $125; M365 Copilot Business $18 to $25.
Press the arrow to build the tiers
Every layer buys capacity. No layer's price includes the operating layer: the people, review, and records that make the capacity governable.
Working language
Portable documentation
Repeatable frameworks
The operating layer is where they live, and someone has to staff it.
Coordination at city scale: Fort Worth's 27 departments, 18 already using AI tools.
Press the arrow to build the three
The buildings are already going up
Part of the code is already written.
Celestica investment figure as reported by Bisnow, May 2026. Building heights are illustrative, not a shared scale.
One code. Different buildings.
Capacity decides how much help an institution needs to reach it.
A low-risk use
Internal drafting from public sources. Named owner. Approved uses. Source checks. Simple records.
A higher-risk use
Resident-facing information or sensitive data, whatever the organization's size. Everything at left, plus review before release, accuracy monitoring, accessibility, and an approval path.
A complex estate
Many systems across many departments. Coordinated inventory, procurement controls, identity management, audit trails, cross-department review.
A checklist becomes infrastructure when it has owners and evidence.
Proposed Human Capacity Infrastructure scaling principle; requirements are use-case and law specific.
Press the arrow to build the three
Who pays to keep the code current?
Texas assigned every covered agency and local government an AI Risk Officer, then declined a proposed shared support platform as unauthorized and unfunded. The duty is assigned. The shared capacity is not.
Funders
Finance shared training, review and evaluation. Avoid unfunded compliance demands.
Public purchasers
Set lawful requirements in RFPs and contracts, verify evidence and name owners.
Chambers and networks
Convene, translate and distribute working materials. Participation remains voluntary.
Independent maintenance and review
Maintain definitions, versioned templates, accessibility, tests and change control. Whoever maintains is never the only one who checks.
Who funds the work, owns the standard and checks the checker?
Potential pathways, not partner commitments or existing legal authority. Counsel review required. Source: Texas Register, adopted rules, March 13, 2026.
Press the arrow to build the four
Fire codes scaled as model codes
Developed centrally. Adapted locally. Adopted by institutions with different duties. No single author finished fire safety, and no single author can finish a capacity standard.
This is a proposed pilot design. The clock starts when a partner owns the workflow and approves the criteria.
Press the arrow to build the rows
An invitation
What I bring
- A named vocabulary and documented frameworks
- A working prototype, GateForce85
- A verification discipline: primary sources first, two independent sources where claims are contested or consequential, claim registers, version authority
- The argument you just heard
What this needs
- Partners with one real workflow
- Funders for shared review hours
- Counsel for the public-sector boundary
- Independent reviewers: no one quality controls their own work, like building an airplane cockpit
- A maintenance home that outlives any one author
Glossary and shared language
Already traveling, not coined here: human in the loop · tool-agnostic · vendor lock-in · operating layer
You do not wait for the fire
to write the code.
Human Capacity Infrastructure is the operating layer between buying AI and institutionalizing it. AI adoption is this deck's application; the discipline is bigger than one technology. Let Tarrant County prove it first: a Fort Worth that invests in human capacity builds a city that serves every resident.
mollywoodbridge.com · GateForce85 · The explainer · No Vendor Gravity · Tool Agnostic
Sources: NIST AI RMF 1.0 · Texas SB 1964 (2025) · 1 TAC Ch. 219 (2026) · Texas HB 149 (2026) · City of Fort Worth Internal Audit 2025.011 · Wistron abatement presentation (2025) · Celestica announcement (2026) · United Way of Tarrant County announcements (2026) · Motorola Annual Report 1988
Band-Aid is a registered trademark of Kenvue Inc., used here as cultural shorthand; no affiliation or endorsement.
Copyright 2026 Molly Woodbridge. All rights reserved. Deck version 3.6.