The Compliance Reality Nobody Tells You About AI

The Compliance Reality Nobody Tells You About AI

Brandon Micci has spent his career inside some of the largest financial institutions in the world, deploying AI and data infrastructure at a scale most companies only talk about. In this episode, we go into what it actually takes to get AI into production inside a regulated organization, where most initiatives stall before they ever reach users, and what the right sequencing looks like when budget and time are limited. We cover the lessons from building a 30,000-user analytics culture at Capital One, the compliance reality of deploying a language model to 27,000 people at JPMorgan, and the practical advice any mid-market company can act on right now to get AI working in the right direction. TIMESTAMPS [0:52] Welcome to The PE Data Guy [1:18] Brandon Micci's Career Background [3:01] Tableau and Capital One Culture [8:25] Safe AI Rollout Best Practices [13:41] Building the CEO Dashboard [17:40] JPMorgan 27,000 User Deployment [22:25] AI Strategy on a Smaller Budget [33:50] Investing with a Long-Term View COMPANIES MENTIONED Capital One, JPMorgan Chase, Citigroup, Southwest Airlines, PwC, Booz Allen Hamilton, Alteryx, Tableau, Oracle, Meta, Palantir, SpaceX, Micron, SanDisk, Lovable, Cursor WEBSITES MENTIONED No specific websites were referenced in this episode. GUEST INFORMATION Brandon Micci is a data and AI executive who has led large-scale initiatives at Capital One, Citigroup, Southwest Airlines, and JPMorgan Chase. At JPMorgan, he oversaw the deployment of an LLM assistant to 27,000 users across the payments organization. He focuses on helping organizations build a clear ROI case before committing resources to AI.

Why 90% of AI Pilots Fail Before They Start

Behind every value creation plan, there's a data problem nobody wants to talk about. In this episode, I sat down with Ben Banks to get into why the vast majority of enterprise AI deployments are delivering no measurable P&L impact, and what AI readiness actually means beneath the slogan. Ben brings 30 years of operational experience, with 18 of those leading global transformations and exit readiness programs across complex PE-backed platforms. The conversation gets real fast. We went deep on business coherence as the actual gating factor holding companies back, not technology. The Zillow Offers story alone is worth the listen. $881 million in losses, 2,000 jobs gone, and every single layer of the transformation stack implicated. We also got into why most AI pilots are structured to fail, what PE firms are missing when they rush to deploy, and how the companies getting AI-ready right now will be the ones buying their collapsed competitors for pennies on the pound by 2027. Timestamps Behind every value creation plan, there's a data problem nobody wants to talk about. In this episode, I sat down with Ben Banks to get into why the vast majority of enterprise AI deployments are delivering no measurable P&L impact, and what AI readiness actually means beneath the slogan. Ben brings 30 years of operational experience, with 18 of those leading global transformations and exit readiness programs across complex PE-backed platforms. The conversation gets real fast. We went deep on business coherence as the actual gating factor holding companies back, not technology. The Zillow Offers story alone is worth the listen. $881 million in losses, 2,000 jobs gone, and every single layer of the transformation stack implicated. We also got into why most AI pilots are structured to fail, what PE firms are missing when they rush to deploy, and how the companies getting AI-ready right now will be the ones buying their collapsed competitors for pennies on the pound by 2027. Timestamps [00:04:27] | AI pilots missing the foundation [00:07:45] | Transformation theater and AI hype [00:09:45] | Workflows vs. replacing humans [00:21:25] | Zillow's $881M AI failure [00:31:50] | Business coherence explained [00:36:36] | Dirty data starts at the top [00:43:37] | AI readiness and exit valuation [00:46:47] | 60-second diagnostic tool Guest Information Ben Banks has spent 30 years inside complex operational systems, with 18 leading global integrations, transformations, and exit readiness programs across tier-one PE funds in the upper middle to large cap market. He built the Growth Engine, an operating framework designed to reduce acquisition lifecycle cost, improve integration velocity, and build cleaner exit stories from day one. He works with PE operating partners and portfolio companies preparing for liquidity events. Companies Mentioned Zillow IBM Capital One McKinsey Forrester MIT OpenAI Anthropic Nvidia Amazon Websites Mentioned https://www.linkedin.com/in/ben-banks [https://www.linkedin.com/in/ben-banks] https://www.zillow.com [https://www.zillow.com] https://www.forrester.com [https://www.forrester.com] https://www.mckinsey.com [https://www.mckinsey.com] Takeaways 90%+ of AI pilots are failing to deliver measurable P&L impact. Getting business coherence right, not technology, is the thing that determines whether enterprise AI works or wrecks a company. PE portfolio companies that are not AI-ready going into a sale in the next 12 months are facing real valuation discounts from buyers who will price in the remediation work. Guest Information Ben Banks has spent 30 years inside complex operational systems, with 18 leading global integrations, transformations, and exit readiness programs across tier-one PE funds in the upper middle to large cap market. He built the Growth Engine, an operating framework designed to reduce acquisition lifecycle cost, improve integration velocity, and build cleaner exit stories from day one. He works with PE operating partners and portfolio companies preparing for liquidity events. Companies Mentioned Zillow IBM Capital One McKinsey Forrester MIT OpenAI Anthropic Nvidia Amazon Websites Mentioned https://www.linkedin.com/in/ben-banks https://www.zillow.com https://www.forrester.com https://www.mckinsey.com Takeaways 90%+ of AI pilots are failing to deliver measurable P&L impact. Getting business coherence right, not technology, is the thing that determines whether enterprise AI works or wrecks a company. PE portfolio companies that are not AI-ready going into a sale in the next 12 months are facing real valuation discounts from buyers who will price in the remediation work.

Why Hiring Works More Like Sales Than HR

We covered why the Rolodex only approach keeps failing, how to write job descriptions that pull in the right people, and what private equity teams can do to build discretionary effort after an acquisition. Frank also shared a simple metric that too many teams ignore: quality of hire, measured early and consistently, not months later when everyone is already frustrated. The theme that stuck with me was clarity. Chapters: * 00:04 - Acquiring Candidates: A Sales Funnel Approach * 04:39 - The Importance of Effective Recruiting * 07:52 - Understanding Onboarding and Probation Periods * 15:56 - The Seven Deadly Sins of Recruiting * 24:53 - The Seven Deadly Recruitment Sins * 32:36 - The Intersection of Dating and Recruitment * 38:30 - Transitioning to Discretionary Effort in Private Equity * 45:42 - Navigating Leadership and Clarity in Private Equity Companies Mentioned IBM Capital One Google LinkedIn Indeed Canva Websites Mentioned tankrecruiting.com [http://tankrecruiting.com] youtube.com [http://youtube.com] Guest Information Frank Lofaro runs Tank Recruiting and helps companies build a recruiting operating system, not just fill a single role. Their process focuses on qualification, amplification, evaluation, and closing so teams can hire faster and reduce reliance on placement fees. Brief Takeaways Quality of hire needs an early signal. Weekly red yellow green feedback prevents surprise failures at month six. Job descriptions work better when they lead with the role’s pull, the results expected, and a clear picture of success.

Why PE Firms Are Asking the Wrong Question About AI

Elon Salfati joined me this week to break down what most PE-backed companies are getting wrong about AI. Not the surface-level stuff, not the board slide version. The real operational gap between clicking a button and actually changing how a business runs. Elon advises PE firms, enterprise operators, and has consulted the UK House of Lords on AI policy alongside leaders from Microsoft and Palantir. His firm, Safari Group, works with businesses to replace manual, people-dependent processes with governed AI systems that scale without adding headcount. What stood out most in this conversation was the idea of decision sovereignty. When a company hands all its strategic thinking to an LLM, it loses its competitive edge. An LLM echoes the past. It will not discover gravity if an apple falls on its server. The real opportunity is flipping the model so the company reaches out to the human for creative judgment, not the other way around. Elon walked through a real case study with Key Loop where they reduced churn by rethinking the entire process workflow before touching the technology. [00:06:43] Wrong question vs right question [00:08:42] Point solutions on broken processes [00:10:39] Political resistance to change [00:16:08] Why AI initiatives stall [00:29:09] Decision sovereignty explained [00:36:09] Human with an army of agents [00:41:19] Turning service companies into software [00:47:01] Personal AI at work and home Guest Information Elon Salfati is the founder of Safari Group, a Zurich-based AI consultancy. His background spans cybersecurity, scalable systems, and applied AI research. He is currently completing his PhD at Imperial College London focused on building secure AI systems and organizational AI culture. Companies Mentioned Safari Group Key Loop IBM Capital One PepsiCo Blackrock NewsCorp Palantir Microsoft Anthropic Websites Mentioned https://www.safari-group.ai [https://www.safari-group.ai] Takeaways AI without process redesign adds chaos, not speed. The companies winning are the ones asking what the operating model looks like when AI runs the operational layer, not just what AI tool to add. Decision sovereignty keeps your competitive edge intact. The shift worth building toward is an army of agents that surfaces decisions to the human, not humans triggering every automation manually.

This Is How He Turned Agency Ownership Into a Video Game

Peter Kang built Barrel into a cash-flowing agency over fifteen years before he ever touched a deal. That patient foundation is what separates Barrel Holdings from a traditional PE firm. No fund, no management fees, no forced exit timeline. Just cash flows reinvested into acquiring good businesses at fair prices, with SBA financing doing the heavy lifting until the flywheel grows strong enough to need less leverage. We talk through what Peter looks for beyond the obvious numbers, including client retention, average tenure, concentration risk, and whether there is real leadership depth below the founder. We get into the Bolster story, the design agency he incubated, loved, and had to walk away from when he realized he was the only reason it had any business at all. And we get into how AI is changing what agencies can deliver, what clients are willing to pay, and how the hourly model is holding on by a thread in a world that rewards output over time. TIMESTAMPS [0:00] Peter Kang and Barrel Holdings [2:02] Holdco Tycoon game origin [5:18] Holdco structure vs PE fund [8:31] Price, leverage, sourcing lessons [16:18] AI reshaping agency work [21:05] What data matters in acquisitions [23:01] Culture risk and Bolster story [30:26] Documenting failures publicly KEY TAKEAWAYS Buying discipline matters more than deal flow. Overpaying is one of the fastest ways to get into trouble, and time pressure only makes it worse. Client retention and average tenure tell you more about an agency than revenue growth. Concentration risk is a red flag worth walking away from. Culture is not a feeling. It is the standards a leadership team enforces and models every day. When those standards clash post-acquisition, good people leave. Publishing failures is a sourcing strategy. Trust is built through transparency, and founders looking to exit remember who showed up honestly. AI is not replacing agencies. It is raising the standard for speed and output, and the pricing models that survive will be the ones tied to value, not hours. COMPANIES MENTIONED Barrel Barrel Holdings Agency Habits Bolster WEBSITES MENTIONED peterkang.com [http://peterkang.com] barrel-holdings.com [http://barrel-holdings.com] https://www.linkedin.com/in/peterkang34/ [https://www.linkedin.com/in/peterkang34/] GUEST INFORMATION Peter Kang co-founded Barrel in 2006 and spent two decades building it into the foundation for Barrel Holdings, a portfolio of digital agencies across e-commerce, Amazon, B2B marketing, and home services. He published The Holdco Guide and created the Holdco Tycoon game to share the capital allocation lessons he learned the hard way. Peter writes openly about wins and failures and runs Barrel Holdings with a decentralized, operator-led model.