NAWIC National Association of
Women in Construction
Findings · Second Congress
2026Construction Congress

The SolutionExists

Reno, Nevada · August 11, 2026

Thirty people who rarely sit at the same table were handed four provocations and told not to agree with any of them. They came back with one diagnosis.

Scroll

01 · The Room

On August 11, 2026, roughly thirty people who rarely sit at the same table sat at the same table.

Ninety seconds inside the room

Play · 1:30

Grand Sierra Resort · Reno, Nevada

Invitation only · 8:00 to 3:00

Who was at the table

Journeypersons and general contractors. Union leadership and corporate risk management.

Apprenticeship administration from the public sector, and workforce development from the nonprofit sector. Safety professionals, manufacturers, insurers, project managers, executives. People with four and five decades in the trades, and people who arrived from other industries entirely.

Several participants entered a workforce that had made no provision for them at all.

The instruction

They were told not to agree.

Four propositions, drawn from what attendees said in advance. The design was adversarial by intent. Each table worked its statement through four timed phases, then had to publish a position it would put its name on.

  1. 10 min

    React

    What in this statement is true, incomplete, or wrong?

  2. 25 min

    Examine

    What is actually happening beneath the headline, and where does responsibility sit?

  3. 25 min

    Challenge

    What would an informed person who disagrees with us say?

  4. 30 min

    Own it

    If this table had to put its name on a one-page position tomorrow, what would we say?

Every table had to leave with five things: a position, no more than three priorities, named owners, one action that could begin immediately, and one question it could not answer.

The day in four numbers

0

Participants,
by invitation

0 of 6

Prepared propositions
that received a table

0

Apprentices lost
in the first year

0 of 4

Afternoon groups that
named culture, independently

Attendance was reduced by a regional wildfire emergency and travel disruption. Two prepared propositions never received a group.

What emerged was not four sets of findings.
It was one.

02 · The Propositions

Four statements.
Four tables.
Four rewrites.

No table left its proposition where it found it. Every one moved the problem further upstream than the statement placed it.

3.1  /  WORKFORCE

“We don’t have a recruitment problem. We have a workforce system problem.”

What the table did to it: Accepted it, then inverted it

The original statement sets recruitment against systems. The table’s version makes recruitment a downstream symptom of a system that leaks. An industry losing half its trainees in year one does not have a recruitment problem. It has an onboarding, supervision, and culture problem that recruitment spending cannot reach.

Owner
Government, education, industry, nonprofit. Industry moves first.
Do now
Form a coalition of subject-matter experts across all four groups.
Unresolved
Money. “We need a lot of money to make this happen.”
  • The missing middle. A layer of mid-career workers left around 2020 and did not return. Senior people are staying past the point they intended to, because there is no one else to teach.
  • Apprenticeship is time-based, not competency-based. A person who has demonstrably mastered a skill still waits out the hours. Against employers offering immediate wages, that is a retention liability, not quality control.
  • Delayed gratification is a luxury good. Unions in one state raised entry wages because warehouse employers were undercutting them. Construction pays far more in five years. A worker living paycheck to paycheck cannot finance a five-year payoff on someone else’s promise.
  • The medicine comparison. Hospital work is hard, unglamorous, and involves the least pleasant parts of human biology, and children still say they want to be doctors and nurses. Medicine also runs the largest apprenticeship system in the country and calls it residency. The obstacle is not the nature of the work. It is the story told about it.

Three priorities: a standing collaboration structure across government, education, industry, and nonprofits; expansion of apprenticeship beyond the trades, given that most merit-shop firms do not use it at all; and marketing that shifts what society counts as a worthwhile life. Children want to be YouTubers because that is the pathway they can see being valued.

3.2  /  TECHNOLOGY AND AI

“Technology will not solve construction’s problems. It will expose them.”

What the table did to it: Refused to resolve it, on purpose

The only table whose irreducible disagreement was the proposition itself. It reported a 50–50 split and named that split as its unresolved question. What it did agree on was the sequence of risk: a workforce without enough construction knowledge to recognise a bad answer is a workforce that will confidently build the wrong thing.

Owner
“Company, government, and guardians.” Parents were named deliberately.
Do now
Build one adoptable AI governance framework using existing industry groups.
Unresolved
The proposition. And the pace: “we don’t know what we don’t know.”
  • Separate the two words. On technology there was little disagreement. Tools have made work measurably safer and faster. A plant station where a worker manually filled bags of concrete and inhaled the dust is exactly the task a robotic arm should take.
  • “AI is a teenager right now.” It learns that item 123 is now item 456. It does not do sarcasm and it does not do critical thinking. One participant described a colleague’s attempt to have it interpret blueprints as an outright failure.
  • What a dashboard cannot see. A person having a bad day is a safety risk, and recognising it requires someone who knows what that person normally looks like. The table cited an incident where a manager saw a worker in acute crisis and intervened in time. If you manage entirely from dashboards, the numbers can look excellent while the thing they describe is failing.
  • The devil’s advocate case, kept in the record. A business owner asked whether a teenage AI is nonetheless better than a workforce that does not show up. Another described AI as genuinely transformative for a small contractor. The counter was equally direct: firms adopting it purely to cut labour cost will hollow out the judgment their business depends on.

Four-part framework: train people on the tools and train the models with the industry’s own expertise; publish a boilerplate governance standard firms can adopt; reframe perception so technology enhances the work rather than substituting for the worker; and educate parents, students, workers, and executives simultaneously, because this one arrived all at once.

3.3  /  SAFETY

“Safety has to move upstream.”

What the table did to it: True, but not sufficient

Their reference was the parable of the river babies. A village pulls drowning children from a river. Some go upstream to find who is throwing them in. But going upstream does not save the people still in the water. Upstream design controls do not relieve anyone of the obligation to the worker who is exposed today. This was the sharpest readout of the day, and it moved safety off process and onto power.

Owner
Everyone, principally those holding position, power, and influence.
Do now
Centre all voices, starting with Indigenous-led and matriarchal models.
Unresolved
The direction of federal policy and the laws that will shape it.
  • One-size-fits-all denies autonomy. Start from the standard operating procedure, then customise to the project and the personnel. Equipment designed for a notional average body on a notional average site produces a hazard for everyone who is not that average.
  • The industry kills its messengers. The safety professional is frequently the most disliked person on the jobsite, for doing the job. And violence here is often not overt. It is delivered at a micro level, and the industry has no accepted vocabulary for it.
  • Women can be part of the problem. The table insisted this stay in the record. A culture of silence is maintained by everyone who participates in it.

The devil’s advocate catalogue, preserved as a field guide to the arguments that stop safety initiatives: we already invested in safety programs; one-off situations don’t represent the company; it’s a budget issue; you knew the job when you took it; it’s always been this way; you’re replaceable; trade workers don’t actually care about process; safety restrictions hinder the work; I’ve never had an issue, so there isn’t one.

Priorities: flatten hierarchy and shift toward community-based and matriarchal decision-making; strengthen accountability at every level and diversify leadership; adopt a speak-your-truth model in place of what the table called the pale-and-stale default.

3.4  /  RETENTION AND CULTURE

“We cannot solve the labor shortage while defending the conditions that drive people away.”

What the table did to it: Rewrote it as a diagnosis

Early on, a participant identified the load-bearing word: defending. The proposition does not accuse the industry of having bad conditions. It accuses the industry of insisting the current state is fine. That distinction organised everything that followed. Their position: the industry is experiencing a culture crisis, and that crisis is driving retention failure.

Owner
Everyone, but “ownership starts at the top.”
Do now
Incentivise conversations about new OSHA and NIOSH regulation.
Unresolved
Transience. Project-based assignments, travel, hours, schedule uncertainty.
  • Hazing apprentices, offered as the first example of a defended condition, justified on the grounds that it is how the people now in charge were taught.
  • The vocabulary problem. With DEI now a contested term, the durable language has become toxic behaviors, which covers more ground than bullying, harassment, and intimidation, and survives political weather. Name the behaviour, not the category.
  • The industry moves the victim, not the aggressor. Raised by a tradeswoman as standard practice during investigations. The person who reported is relocated.
  • Accountability is credible only when it is expensive. The table’s benchmark was a large general contractor known to shut down a jobsite at enormous cost when an incident occurs and no one takes responsibility.
  • Transience taxes women twice. A tradeswoman may work for twenty contractors in a year. Camaraderie requires working with the same people repeatedly. And each new site restarts the burden of proof: a woman arrives having to demonstrate skill, a man arrives presumed to have it.
  • Corporate is not exempt. The office has cubicle farms, bad managers, and unworkable policies too. It is also where the fix must originate, because the problem is not that corporate cannot solve it but that corporate frequently declines to acknowledge it exists.
  • Family care as a scheduling problem. Start times that precede daycare opening hours are a retention decision disguised as a schedule.

A concrete policy ask: that OSHA establish a standard covering mental health, expanding the Focus Four to include psychological safety, and that the standard also address toxic behaviours.

“There is no incentive in this industry to be truthful about safety. There is a reward for silence.”

Safety table · midday readout

The most actionable finding of the day

Some companies assert that they answer to U.S. law and not to tribal law.

That position functionally removes safety protection from Indigenous workers on those projects. The consequences the table described were fatal ones, in cases that too often close without resolution.

It is currently invisible in mainstream industry safety discourse. It is specific, it is documentable, and it is the single item from this Congress that could be acted on tomorrow.

03 · The Convergence

The afternoon was designed to break the morning’s conclusions.

Tables were remixed so that everyone sat with people who had spent the morning on a different problem. Then they were handed positions they had not written and asked to defend them.

The first question of the afternoon was simple. Which of these issues are actually parts of the same problem?

Fig. 04 · One ground

AFTERNOON GROUP A Culture and system. Safety, retention, knowledge transfer, nepotism. AFTERNOON GROUP B Where everything connected. And everything came from the top down. AFTERNOON GROUP C Perception and communication, all of it filed under one heading. AFTERNOON GROUP D Perception, culture, communication, with accountability as the baseline. ONE ROOT CAUSE Named independently, 4 of 4
  • Group A Culture and system. Safety, retention, knowledge transfer, and nepotism all connected to it.
  • Group B Culture is where everything connected, and everything came from the top down.
  • Group C Perception and the failure of communication, then all of it filed under one heading.
  • Group D Perception, culture, and communication, with accountability as the baseline.

Four groups. Four different problems. One word.

Culture

Why this one counts

Four differently constituted groups, working from positions they did not author, converged on one root cause.

That is not a facilitated consensus. That is replication.

And the warning that comes with it

“Culture” is the most abused word in industry reporting. It becomes a container for everything and therefore a mandate for nothing.

What protects the finding here is that every group also named the mechanism. Leadership behaviour, which is observable, assignable, trainable, and firable.

The room did not say it needs better culture. The room said, repeatedly and from four directions, that culture is set by what leaders reward and tolerate, and that the industry’s leaders are not currently doing the work.

“We think everything is a leadership issue.”

Afternoon group · readout

04 · Cross-Cutting Findings

Six things that were true at every table.

Finding 01

The industry habitually intervenes at the point where a problem becomes visible, rather than where it becomes irreversible.

Recruitment became system. Safety became standing. Technology became judgment. Retention became leadership. Every table moved its problem earlier in the causal chain than the proposition placed it.

Recruitment campaigns are the clearest case. They are the most fundable, most measurable, most photogenic intervention available, and they operate at the exact point where the room’s evidence says leverage is lowest.

Finding 02

Leadership is the named mechanism, and the room does not think leaders are equipped.

One afternoon group proposed transformational leadership training as the single highest-leverage intervention available. If everything is a leadership issue, then leadership capability is the constraint.

This is more demanding than it sounds. It implies that the industry’s leadership development, where it exists, is not producing people who can do this. A participant asked the corresponding question and did not get a satisfying answer: how do we incentivise supervisors and managers to actually get better, and who pays for it? The room named the requirement. It did not solve the incentive.

Finding 03

Retention is the leverage point, and the industry measures the wrong end.

The industry measures applicants and hires. The failure is concentrated in the first year of apprenticeship, in supervisory quality, and in whether a worker would recommend this industry to someone they care about.

An industry that loses half its apprentices in year one is running a very large, very expensive, entirely negative word-of-mouth campaign among exactly the demographic it is trying to recruit.

Retention is not the reward for good marketing.
Retention is the marketing.

Finding 04

The industry has evidence it does not believe, and lacks evidence where it most needs it.

Studies on apprenticeship retention, inclusion, and culture already exist and have not changed behaviour. Meanwhile there are no trusted baseline statistics on jobsite violence, harassment, and bullying that would let the industry see itself in perspective against other sectors.

The group asking for that data was notably free of self-pity. They wanted the comparison not to indict the industry but to be able to talk about it honestly, including the possibility that construction looks better on some measures than its reputation suggests. In their words: it is not perfect, so let us not gloss over it, but it is not as bad as people think, so let us get real.

Finding 05

Authenticity is now a competitive requirement in industry marketing, and the industry is failing it.

A participant described marketing teams dressing office staff in PPE for jobsite photography, and what happens when a photo gets caught out over a detail as small as visible jewellery. The audience does not believe you, and it tells everyone.

The observation underneath is the useful part. Construction marketing departments are staffed by people who studied marketing, not construction. They are asked to represent a workplace they have never worked in, and they are given no authentic assets, because jobsites restrict cameras for liability and proprietary reasons.

The result is staged imagery that the industry’s own workforce recognises as false. Every recruitment dollar spent on it is spent on a message its target audience has already learned to discount.

Finding 06

An afternoon table arrived at nepotism as a root cause of the talent shortage. Then it stopped.

Someone had noticed something. It had not come up in any morning conversation. They checked around the table, with participants drawn from all four morning groups, and confirmed it.

It had come up nowhere. In a full day convened specifically to examine how people enter this industry, no table had raised it once.

“It’s everywhere.”

Their own reaction, on realising it had gone unnamed all day

What followed was the most specific testimony of the day

Intake interviews ask whether an applicant has fixed a bike, changed a tire, or worked on a car with a father or grandfather. Those questions are designed to test aptitude.

They function as a proxy for who had access to tools, and to a man willing to teach them.

Which is, structurally, a screen against women and against anyone without that household. A participant who sits on an apprenticeship intake committee described candidates advanced as a political responsibility, the son or grandson of a member, and described how the term is used knowingly inside the process.

Test for the ability to learn rather than for what the candidate already knows, on the grounds that an apprentice is by definition someone who knows nothing yet. The value they wanted to select for was grit and hunger, which they argued is the one thing that cannot be trained.

Their objection to nepotism was not primarily fairness. It was quality. The industry is systematically selecting for connection over capacity while complaining about a talent shortage.

And there is no symmetry across the country. Direct-entry sponsorship is routine in one state and grounds for a reprimand in another, with a competitive tryout day used elsewhere. The same national organisations run materially different systems. Symmetry of process belongs in any published recommendation, because it is concrete, it is inside the industry’s control, and it requires no external funding.

05 · Where the Room Did Not Agree

The facilitation was built to preserve disagreement, not resolve it.

When a table agreed too quickly, it was told that speed of agreement usually means the idea has not been tested. These tensions are live. Each one is a place where a recommendation could fail on contact with the industry.

Five live tensions

Inclusion versus exclusion

In trying to be inclusive, the industry risks becoming exclusive, alienating advocates who are not members of the group being centred.

There are men in this room who are advocates for women in construction. A strategy that alienates them costs more than it gains.

Whose problem is this?

Asked directly, early in the morning: are we answering these questions for women in the industry, or for the entire industry regardless of gender?

The facilitator ruled: industry, with different lenses applied. That ruling shaped every position produced that day and should be stated openly rather than left for readers to assume.

Is “they’ll leave anyway” a constraint or an excuse?

Employers will not invest in training people who leave. Most of the industry is small businesses that may lack not the will but the capital and the starting point.

A working contractor pushed back from direct experience: transience is the nature of the work, crews are hired and released as jobs demand, and a hall of untrained people produces worse crews on the next build.

Upstream versus downstream safety

Going upstream to find the cause is the only way to stop the harm at source.

One participant reported having been alone on the other side of the river, arguing the upstream case against the table’s downstream amendment. The final position preserved both, and it should stay preserved.

Time-based versus competency-based advancement

The efficiency argument is obvious. A person who has mastered the skill should not wait out the hours.

The objection, that hours in the field teach things a competency test does not measure, was gestured at but never fully made. It deserves a dedicated session at the next Congress.

The most consequential open question in the proceeding

Mandates and penalties, or incentives?

Case one

A major Boston project with binding numerical targets for women apprentices and journey-level workers. An on-site ombudsman funded through a third-party nonprofit who counted every day. Real penalties for missing targets. A contractor group went beyond the requirement rather than paying the fine.

Conditions: a supportive mayor, an organised policy coalition, a union hall next to the site, sustained advocacy.

Case two

Other major projects where contractors simply budgeted the penalty and absorbed it as a cost of doing business.

The room split. One position: penalties must be high enough that compliance is cheaper than the fine. The other: incentives and enhanced payment work better than punishment.

Neither side produced evidence that would settle it. The answer determines whether the industry’s reform pathway runs through procurement and policy, or through voluntary leadership.

06 · What the Room Could Not Resolve

Each table had to name one thing it could not settle.

TableUnresolved issue
WorkforceFunding. The coalition and three-year plan require money nobody at the table could source.
Technology & AIThe proposition itself, split 50–50. And the pace of change: “we don’t know what we don’t know.”
SafetyThe direction of federal administration policy and the laws that will shape it.
Retention & CultureThe transient nature of the work. Project-based assignments, travel, long hours, schedule uncertainty.

Three of the four are outside the industry’s control.

That is an uncomfortable result for a Congress convened on the premise of what the industry can own, and it is reported here rather than resolved editorially.

But the fourth is not external. The transient nature of construction work is a design choice embedded in how projects are procured, staffed, and scheduled. It is difficult, it is deeply structural, and it is inside the industry. It is precisely the right subject for the next Congress.

07 · The Empty Chairs

The room named its own missing seats.

Two afternoon groups made the composition of the Congress part of their findings. That is a mark of an unusually self-aware convening, and it is the most awkward thing in this document.

Who was not at the table

  • Educators

    No educators were present, in a room whose single most repeated recommendation was collaboration with the education system. Educators were part of the 2025 Congress. This is a regression, and participants flagged it themselves.

  • Owners and capital

    The people who control project money were not present. Several participants held leadership positions. None represented the client side that ultimately sets conditions through procurement.

  • Students and early-career workers

    A group asked directly for people two years into the industry. The entire retention analysis rests on understanding why people leave in year one, and nobody in year one was in the room.

  • A fuller demographic range

    A group named the absence of all races and age groups as a gap in its own evidence base.

  • More trades voices

    Named as a requirement for honest discussion, despite significant trades representation already being present.

The conclusion is not that the Congress was badly composed. It is that the room correctly identified that its recommendations are addressed to parties who were never asked.

Two propositions never got a table

Knowledge transfer showed up anyway. Everywhere.

Without a table and without a prompt, it recurred in nearly every conversation. The missing mid-career layer. Senior workers staying past their intended retirement because there is nobody to teach. Regions now recalling retired workers to solve problems precisely because transfer has failed. A proposal to use AI to capture standard operating procedures from senior staff before they leave.

One participant framed the ambition better than a prompt would have: the goal is not to transfer knowledge but to accelerate it. Young people master enormously complex game systems in weeks through iterative failure. Why can project management judgment not be taught the same way?

That it surfaced this insistently without being asked is itself the finding. It is not a separate industry issue. It is the mechanism by which the culture problem and the workforce problem are the same problem.

The other one

Responsible growth did not surface at all.

In roughly four hours of recorded proceedings, the question of who bears the workforce, infrastructure, environmental, and community costs of major development was effectively absent. Data centers were mentioned once, in passing, and only in the context of photography restrictions.

This is named here rather than quietly omitted, because the omission is the point. The industry is in the early stages of a construction boom in energy, semiconductors, advanced manufacturing, and data infrastructure. A room of thirty experienced professionals, convened specifically to examine the industry’s hardest problems, did not spontaneously raise the question of what that boom does to communities.

That is not a criticism of the participants. They were given four other propositions and ninety minutes. It is a finding about where the industry’s attention currently sits.

08 · Priorities for the Next Five Years

Four groups produced twelve priorities.
Three survive all four lists.

Priority one

Make the business case for culture in the currency owners use.

This is the enabling priority. Without it, every other recommendation competes for discretionary spending against work that has a bid number attached, and it loses that competition every time, on the merits, in every firm.

The deliverable

A defensible ROI framework linking culture, supervision quality, and retention to project cost. Attrition and replacement, rework, schedule impact, insurance and claims exposure. Sized for firms that are small businesses, not for the top hundred contractors.

The honest version of the problem is not that contractors are unaware. So much of the industry is small businesses operating at the margin of their capacity, barely able to keep work moving, let alone invest in a program whose payback is diffuse and slow. The case has never been made in the currency they operate in.

Priority two

Treat leadership capability as the binding constraint, and build for it.

Every table located the mechanism in leadership behaviour. The named deliverable was transformational leadership training. But the room also identified the missing half.

The distinction that matters

Leadership training without an incentive to attend is a line item. Leadership training tied to advancement, contract eligibility, or insurance treatment is a lever.

Priority three

Fix the front door, and the honesty of the industry’s self-portrait.

This consolidates the workforce, perception, and nepotism findings into one addressable program. Every element is inside the industry’s control, and none of it requires legislation.

  • Competency-based progression alongside time served.
  • Symmetry of entry process across regions and locals, so the same national organisations stop running materially different systems.
  • Intake criteria that test the capacity to learn rather than prior access to tools and to someone willing to teach.
  • Authentic, freely available imagery of real people doing real work, so the industry stops advertising itself with pictures its own workforce recognises as false.

Two actions that require no consensus and can begin immediately

Commission the missing data.

A comparative baseline on jobsite violence, harassment, and bullying against other sectors, and a credible earnings-trajectory analysis. The room named both as gaps. Neither requires agreement to start. Whichever organisation commissions and publishes that work will occupy a position no one in the industry currently holds.

Build the coalition, and staff it.

Named independently by two morning tables and echoed in the afternoon. The room’s own warning should be written into its charter: a task force that convenes the already-convened is worse than none, and a genuine one requires the people currently outside the room, including the people the industry finds difficult.

09 · The Elephant in the Room

Every table did it, and no table examined it.

Asked who owns the problem, every group said everyone.

And in an industry where everyone owns a problem, no one does.

The groups sensed it and corrected themselves within a sentence. Everyone, but it starts at the top. Which is closer to true, and still leaves the question open.

Because

  • The people at the top of the industry were not in the room.
  • The owners were not in the room.
  • The holders of capital were not in the room.
  • The education system was not in the room.

The room said so itself. Twice. Unprompted.

The construction industry has developed a habit of convening the people who experience its conditions rather than the people who set them.

An integrator that convenes only its own membership is a caucus, not a coalition.

The first test of the mandate

Three findings only make sense once the elephant is visible

Why good evidence does not move the market

The studies exist and the industry does not believe them. The more precise explanation is that the evidence has never been addressed to the party that decides. Research written for practitioners does not reach the owner setting the schedule, the client selecting on lowest bid, or the insurer pricing the risk. The case has been made repeatedly in the wrong currency, to the wrong room.

Why penalties get budgeted

The difference between the project that exceeded its targets and the projects that absorbed the fine was not contractor virtue. It was whether the owner, the public authority, and an independent monitor were actually engaged. Where the parties who set conditions participated, the targets held. Where they did not, the penalty became a price.

Why “everyone owns it” keeps producing nothing

Diffuse ownership is not a moral failing. It is the predictable output of a fragmented delivery model in which no single actor controls enough of the system to be held responsible for it. General contractors do not control owners. Owners do not control the labour supply. Unions do not control merit-shop firms. Educators do not know what any of them need. Every party can accurately say the problem began somewhere else, and every party is right.

What follows from naming it

The corrective is not another convening of the willing. It is an integrator.

Individual firms cannot act alone on culture, apprenticeship symmetry, or data collection without absorbing a competitive cost for doing so. Membership organisations can, because they exist above the level where that competition happens.

But the mandate comes with three tests, and they are demanding.

  1. 01

    Composition

    An organisation that convenes only its own membership has held a meeting, not built a coalition. The credibility of everything in this report depends on who is in the room next time. Specifically educators, owners and capital, and workers in their first two years.

  2. 02

    Evidence

    The room asked for data it does not have and does not trust others to produce impartially. Whichever organisation commissions and publishes that work occupies a position no one currently holds.

  3. 03

    The tension the room did not resolve

    Any integrator must centre the constituency it exists to serve while keeping the advocates who sit outside it. This will be the hardest part of the work, and pretending otherwise is how coalitions fail.

10 · What the Industry Does Next

What the industry lacks is not insight.
It is arrangement.

Thirty experienced professionals, given four different problems and explicit instructions to disagree, produced one diagnosis. That is not the normal state of an industry’s self-knowledge. Most sectors argue about the diagnosis for a decade before they begin arguing about the remedy.

Three things follow, and none of them require consensus to begin.

Move one

Stop funding the front door. Start fixing the first year.

Recruitment is the most fundable, most measurable, most photogenic intervention available, and the room’s evidence places it at the point of lowest leverage.

Move two

Make the case in the currency of the people who decide.

Culture, supervision quality, and turnover have measurable cost consequences. Until that arithmetic is assembled and sized for the small businesses that constitute most of the industry, every recommendation here competes against work that has a bid number attached.

Move three

Fix what is entirely inside the industry’s own control.

Competency-based progression. Symmetry of entry across regions and locals. Intake that tests the capacity to learn. Authentic imagery of real people doing real work.

Every one of these is achievable without legislation, funding, or agreement from anyone outside the industry.

The closing assessment

“The solution exists. We’re just so fragmented we can’t pull it off.”

Participant · midday plenary

Fragmentation is not a mystery and it is not a character flaw. It is a structure. A delivery model that distributes responsibility so thoroughly that every participant can truthfully say the problem began somewhere else.

Structures can be changed, but only by actors positioned above the level where the fragmentation occurs, and only when someone accepts a mandate that no single firm has an incentive to accept alone.

That is the work.
The industry already knows what to do.

Back matter · how this was made and where it is weaker

Proceedings were captured on multiple audio sources and transcribed by automated speech recognition, covering roughly four hours of the working day. Two overlapping capture passes exist for portions of the proceedings, recorded by different devices at different distances from the speaker. That redundancy was used as a verification instrument: where the same passage exists in two independent transcriptions, discrepancies reveal transcription error rather than speaker inconsistency, and the clearer capture was preferred.

The evidence standard applied:

  • Formal readouts were weighted highest. Every table position, priority, owner, action, and unresolved issue in this report derives from a readout delivered on microphone.
  • Small-group discussion was used for evidence and texture, never for positions.
  • Speaker diarization was not used for attribution. Automated speaker labels split single speakers across multiple labels and merge different speakers under one. Attribution here is by role and context.
  • Where audio was unclear, the claim was excluded rather than reconstructed. Several interesting exchanges were dropped for this reason.
  • Direct quotation is used only where the audio is clean and the phrasing unambiguous. Everything else is paraphrased.

Known limitations. Three of the four morning tables have partial or minimal coverage in the record. The positions are secure because they were presented at midday, but the reasoning behind some positions is thinner in the record than it was in the room. The Congress had four active tables, not six.

Anonymity standard. No participant is described in enough biographical detail to be identified by search. Positions are attributed to tables and to broad professional roles. Career details, organisational statistics, tenure figures, and regional identifiers have been generalised or removed, including where they were compelling. Several vivid passages were sacrificed to this standard.

The following were made in the room and are load-bearing enough that they should be independently sourced before wider publication. None is disputed here. All are flagged because industry findings should not rest on a conference-floor citation.

  • Roughly 50% of apprentices lost in the first year, across trades. This figure anchors the workforce argument. Source it to a specific dataset with the population defined.
  • Approximately 714,000 registered apprentices nationally, women at 5.4% for FY2025. Directionally consistent with published federal data. Confirm whether the 5.4% figure is construction-specific or all-industry.
  • A proposal that OSHA establish a standard on mental health and toxic behaviors, expanding the Focus Four. Cite the referenced work directly.
  • The Boston project with binding participation targets, an on-site ombudsman, and enforced penalties. The strongest counter-example to the penalties-get-budgeted problem. Confirm project, dates, target, and result, and give it a proper case study.
  • A major project where contractors budgeted and absorbed the penalty. The pattern can be reported without identifying the project.
  • A large general contractor’s practice of shutting down a jobsite when an incident occurs and no one takes responsibility. Do not name the firm without confirmation. The principle stands without the attribution.
  • Licensing cost cited in the low thousands for a single industry safety video. Verify or generalise. The structural point holds regardless.

On the industry’s condition. “The solution exists. We’re just so fragmented we can’t pull it off.” · “The industry is experiencing a culture crisis.” · “We are the change we’ve been looking for.”

On leadership. “Everyone owns it. It’s leadership.” · “Supervision sets the tone of the job.” · “Ownership starts at the top.”

On safety. “Culture and safety are not two separate issues.” · “We generally move the victim rather than the aggressor.”

On entry and access. “I can work with grit. I can’t work with a lack of ambition.” · “Why aren’t we testing for what your ability to learn is?”

On perception. “It’s not perfect, so let’s not gloss over it, but it’s not as bad as people think it is. So let’s get real.” · “The dirty laundry is going to have to get aired out eventually.”

On technology. “AI is a teenager right now.” · “We don’t know what we don’t know.” · “A robot can never replace the human mind.”

On change. “That’s the way it’s always been done. I was taught that way, so you’re going to learn that way,” offered as the thing to defeat. The rebuttal: we all embraced the dishwasher and the power window without argument.