The Real AI Divide in Houston Is Not Access. It’s Execution.

Executive Summary

• In this Houston workshop dataset, leaders most often want AI for revenue growth (71%), followed by cost reduction (18%).

• The biggest blocker is talent and skills (47%), with tech stack/tools (18%) and regulation/compliance (18%) tied for second.

• The ownership picture is mixed: 29% report no clear owner, 12% rely on a working group without a single owner, 29% name a functional leader, and 29% say the CEO/GM is accountable.

• Delivery maturity is still early: 53% say zero pilots have made it into production in the last 12 months, and 65% have no AI KPIs yet.

• Data readiness is uneven. Only 6% report having a clean, labeled dataset with access controls ready for a 30-day pilot.

• Confidence is moderate at 6.6/10 on average, but respondents with at least one use case tied to a KPI, named owner, and review cadence report materially higher confidence than those with no KPIs.

• External industry research points in the same direction: most sectors are now beyond basic experimentation, and the organizations creating real value are the ones that pair AI adoption with workflow redesign, leadership ownership, data standards, and governance.

What the Survey Reveals About AI Readiness

Outcomes leaders want

The Houston survey is clear about intent. Leaders are not asking AI to be interesting. They are asking it to be useful. The dominant outcome is revenue growth, chosen by 12 of 17 respondents. Cost reduction is a distant second, and only one respondent each named customer experience or closing a talent gap as the top goal.

That matters because it tells you where AI will be judged. In this group, AI is not being framed mainly as a research tool or innovation lab. It is being evaluated as a commercial lever. Leaders want faster growth, cleaner follow-up, better quoting, sharper execution, and more productive use of existing teams.

What’s blocking progress

The main obstacle is not lack of awareness. It is lack of capability. Nearly half of respondents chose talent and skills as the primary blocker. Tech stack/tools and regulation/compliance came next, while budget and data quality showed up less often as the single biggest barrier.

There is a useful nuance here. Teams often talk about tooling first, but the Houston responses suggest that many leaders already know the harder part is internal: who can lead this work, translate it into operations, and keep it moving after the workshop ends. One anonymized comment captured that well: a finance leader noted they had already adopted third-party tools to improve processes, but lacked the time to build deeper internal capability.

The ownership gap and why it matters

The ownership pattern is what separates curiosity from execution. In Houston, 7 of 17 respondents either have no clear owner or rely on a working group without a single accountable person. That means 41% of the sample still lacks one point of accountability.

The downstream effects show up quickly. More than half of respondents say no pilots have made it into production in the last year. Nearly two-thirds say they do not have KPIs tied to AI yet. Most teams also move slowly when performance drops: 59% say they can only make a production change within a month or a quarter, while just 12% say same day.

The pattern inside the data is telling. Respondents with at least one AI use case tied to a KPI, named owner, and regular review cadence reported an average confidence score of 8.6/10. Those with no KPIs averaged 5.8/10. The survey does not prove causation, but it does point to a practical truth: measurement and ownership tend to travel together.

Industry Intelligence: How 5 Sectors Are Responding to AI Right Now

Construction and infrastructure

Construction is moving from document digitization to execution support. The pressure is not theoretical. McKinsey reports that construction productivity improved only 10% total, or 0.4% annually, from 2000 to 2022, versus much faster gains in manufacturing and the broader economy. Deloitte’s 2025 construction adoption study found that 37% of construction businesses are using AI and machine learning, and the average firm now uses 6.2 technologies, up from 5.3 the year before.

The realistic AI use cases are schedule risk prediction, RFI and submittal intelligence, estimating support, safety documentation, and project controls. The common pitfall is the same one visible in the Houston survey: tools get added, but workflows do not change. Construction’s weak spot has never been lack of software. It has been inconsistent adoption at the project level and difficulty scaling improvements across the portfolio.

Energy, petrochemicals, and subsea operations

In energy and process industries, the center of gravity is shifting from isolated digital projects to frontline operational use. Deloitte’s 2026 oil and gas outlook says around half of all AI and generative AI spending by U.S. oil and gas companies now targets process optimization. The same report cites an example where predictive algorithms prevented more than 140 hours of downtime and protected 1.6% uptime. McKinsey notes that energy and materials are especially well positioned to benefit because these sectors already run on complex operational data and analytics.

That makes the Houston responses from petrochemical and subsea-linked firms unsurprising. The likely use cases are maintenance, optimization, troubleshooting, throughput, and risk management. The biggest pitfalls are fragmented operational data, heavy dependence on specialist knowledge, and slow change cycles. In sectors where one decision can affect safety, uptime, and margin all at once, AI only matters if it is tied to governed operational routines.

Industrial technology, instrumentation, and integration services

Direct public research on instrumentation and integration firms is thinner than for large manufacturers, so the best proxy is smart manufacturing and maintenance research. Deloitte’s 2025 smart manufacturing survey found that 29% of manufacturers are already using AI/ML at the facility or network level, 24% have deployed generative AI at that same level, and 54% report using a data standard through a unified data model. McKinsey’s maintenance research adds a practical view from the field: one industrial gen AI copilot reduced unscheduled downtime by as much as 90%, cut maintenance labor costs by a third, and increased technician capacity by 40%.

For Houston firms in instrumentation, tech, and integration services, the realistic applications are maintenance copilots, documentation retrieval, troubleshooting support, field service knowledge capture, and workflow automation. The common pitfall is assuming that strong technical talent alone is enough. In practice, these environments still need data standards, change management, and clear review rules before pilots can become repeatable operating tools.

Financial services and lending

Financial services is one of the most heavily invested AI sectors. The World Economic Forum reports that financial services firms spent $35 billion on AI in 2023, with projected investment rising to $97 billion by 2027. The same report notes that 32% to 39% of work across banking, insurance, and capital markets has high automation potential, with another 34% to 37% holding high augmentation potential. Gartner found that 59% of finance leaders reported using AI in the finance function in 2025, with knowledge management, accounts payable automation, and anomaly detection among the most common use cases.

That external reality makes the Houston banking and mortgage responses especially instructive. These firms are not behind because the sector lacks use cases. They are behind when data, governance, and ownership are weak. Deloitte’s financial services research shows that pioneers with stronger expertise and operating discipline are more likely to exceed ROI expectations and achieve their intended benefits. The lesson for lenders is straightforward: the issue is no longer whether AI belongs in finance, but whether the institution can operationalize it safely and measurably.

Marketing and customer acquisition services

Marketing is one of the clearest examples of AI shifting from optional to structural. Gartner says martech utilization has dropped to 49%, which means many organizations are layering AI onto stacks they are not fully using. In a separate 2025 survey, Gartner found that 81% of marketing technology leaders were either piloting or already implementing AI agents, but half said they lacked the technical and data-stack readiness required for deployment. McKinsey adds that roughly 50% of Google searches already show AI summaries, and that figure is expected to exceed 75% by 2028.

For Houston participants working in marketing and dealership services, that changes the operating environment immediately. The realistic use cases are content production, campaign optimization, lead scoring, account prioritization, and search visibility adaptation. The biggest pitfall is mistaking content output for commercial performance. AI can help produce more, faster, but the firms that win will be the ones that connect AI to conversion, pipeline quality, and discoverability across AI-mediated search and decision journeys.

What High-Performing Organizations Are Doing Differently

Across industries, the playbook is becoming more consistent. McKinsey’s 2025 global AI survey found that most organizations are still in experimentation, but high performers stand out in a few specific ways: they redesign workflows, secure visible senior leadership ownership, define when human validation is required, and track KPIs tied to business outcomes. They are also more likely to pursue growth and innovation, not just cost reduction.

That maps neatly to the Houston data. The firms that feel more confident are not simply “more interested in AI.” They are more likely to have an owner, a KPI, some review cadence, and a cleaner connection between pilot and production.

Recommendations Informed by the Workshop Data

First, name one accountable AI owner for the next 90 days. That directly addresses the 41% of respondents who still lack a single accountable person.

Second, build a KPI scoreboard before launching the next pilot. Houston’s biggest execution problem is not imagination, it is lack of measurement. Pick a small set of metrics tied to the goal leaders actually named: revenue, cost, response speed, or customer experience.

Third, start with one workflow, not one tool. The data shows raw or siloed data is far more common than clean, ready datasets. Choose one process where the business case is visible and the data can be made usable.

Fourth, treat governance as an enabler. While 29% report strong blocking and review rules, the rest rely on informal habits, partial enforcement, or no protections at all. That is manageable at pilot stage and dangerous at scale.

Fifth, build skills around the workflow, not around the platform. Talent and skills were the top blocker. The fastest way to close that gap is to teach teams how to run one valuable use case end to end.

Sixth, define a pilot-to-production gate. If a use case lacks a named owner, KPI baseline, review process, and safe-use rules, it is not ready to move beyond experimentation.

For organizations that want to turn the survey’s strong interest in internal AI leadership into action, a practical next step is to explore a leadership-development path such as GPS Summit enrollment, review how GPS Summit compares with university programs, or learn more about BREATHE! Exp.