Engineering Alpha in Private Equity

Paul Karner and Dave Mangot

Engineering Alpha in Private Equity is a podcast about how software engineering and data science excellence create operational alpha. Hosted by Dave Mangot, the author of _DevOps Patterns for Private Equity_, who has worked with operating teams at Thoma Bravo, Hg Capital, and other top-tier PE firms. Co-hosted by Paul Karner, PhD, an economist with two decades inside PE-backed companies. Each episode explores the intersection of technology decisions and investment outcomes.

  1. 13h ago

    The On-Demand Patching Trap: Unreleased Inventory, Due Diligence Audits, and the Real Foundation of AI Velocity

    When software portfolio companies struggle with production outages or fail to capture ROI from AI-generated code, Operating Partners often look at engineering throughput, code review bottlenecks, or team sizing. But what if the real bottleneck isn't feature delivery at all? What if it's the hidden operational friction of how the company patches and upgrades its core infrastructure? In Episode 23 Paul and Dave unpack why legacy "quarterly patch schedules" and maintenance windows are actively destroying portfolio value. Prompted by recent warnings from frontier AI security leaders about a coming "vulnerability tsunami," Dave and Paul explain why you cannot layer AI code generation onto a deployment pipeline terrified of production. They connect automated CI testing and immutable deployments directly to exit multiples, cyber insurance insurability, and customer churn—giving operating partners a blueprint for turning critical updates into an operational "nothing-burger". Key Takeaways: - The "Quarterly Patch" Fallacy: In a world where dependencies and vulnerabilities surface daily, waiting for quarterly maintenance windows doesn't reduce risk. It builds up dangerous unreleased inventory that triggers longer outages and customer attrition. - The AI Delivery Paradox: If AI agents write 10,000 lines of code a day, but your deployment pipeline can only safely release code once a month or once a quarter, AI investment will only accelerate production failures rather than EBITDA. - Driving Out Fear with Automated Testing: Teams delay patching and upgrades not out of negligence, but fear: “If I patch this, will it break production?” The only way to eliminate that fear is comprehensive, automated CI test suites that validate dependencies before deployment. - Due Diligence & Cyber Insurance Impact: Sophisticated buyers audit software supply chains and release cadences during technical diligence. Companies unable to prove automated, on-demand remediation face higher insurance premiums, steep valuation discounts, or deal-closing escrow holdbacks. - The Compounding Return on Basics: Investing in automated test harnesses and continuous deployment pays off from multiple angles: protecting the balance sheet from external disruption while unlocking the foundation needed to actually capture AI returns.

    The On-Demand Patching Trap: Unreleased Inventory, Due Diligence Audits, and the Real Foundation of AI Velocity
  2. Nov 10

    The Analog-to-Digital Playbook: Frontline AI, The Goldilocks Target Zone, and Operational EBITDA

    With macroeconomic headwinds ending the era of cheap leverage and rising-tide multiple expansion, Private Equity firms can no longer rely on financial engineering alone. In Episode 22 host Dave Mangot sits down with Cory Eaves, Partner and Head of Portfolio Operations at BayPine (former 3-time portfolio company CTO and 15-year veteran at General Atlantic), to unpack how hands-on operational transformation drives real EBITDA expansion across traditional businesses. Drawing on his background as an engineering leader and operating partner, Cory outlines BayPine’s "Analog to Digital" investment strategy across healthcare, industrials, and financial services—the 85% of the global economy sitting outside the traditional tech sector. He details how BayPine evaluates target readiness using their "Goldilocks Zone," balances top-down process re-engineering with bottom-up frontline AI adoption (including a firm-wide OpenAI deployment that reduced factory machine downtime from 4 days to minutes), and automates administrative toil to redeploy staff toward high-touch customer retention. Key Takeaways: - The "Analog to Digital" Thesis (85% of the Economy): While tech companies comprise ~15% of the global economy, the remaining 85% (industrials, healthcare, business services) still runs on legacy paper processes and PDFs. Modernizing these non-tech businesses with software, data, and AI represents a massive margin expansion opportunity in middle-market PE. - The "Goldilocks Zone" for Target Selection: Ideal acquisition targets aren't digital winners (which don't need operational help) or digital deserts (where the technical lift is too heavy). Operating Partners should look for companies with "good basic bones"—core systems and organized data ready to be unlocked. - Bottom-Up AI Adoption vs. Top-Down Overhead: Rather than forcing massive top-down IT mandates, BayPine deployed thousands of OpenAI seats directly to frontline workers. In one manufacturing PortCo, plant workers used smartphones to diagnose milling machine error codes and execute repairs on the spot—slashing expensive machine downtime from 4 days to minutes without added OPEX. - Addition Over Subtraction (Redeploying Capacity): In most cases, AI adoption in non-tech PortCos should be framed around employees reskilling or redistribution versus purely as headcount reduction. By automating tedious administrative toil (such as manual transcript verification in education), staff are redeployed to high-touch, value-adding student support that drives top-line retention.

    The Analog-to-Digital Playbook: Frontline AI, The Goldilocks Target Zone, and Operational EBITDA
  3. Oct 29

    AI-Washing Layoffs, Budget Reallocation, and the Valuation Multiple Trap

    When major tech firms announce massive workforce reductions while simultaneously pouring billions into AI infrastructure, PE board members and Operating Partners are left asking: Are these layoffs driven by true AI productivity gains, or are companies simply reallocating budget to pay for AI CapEx? In Episode 21 Paul and Dave apply an economic lens to AI-era labor market disruptions. They deconstruct the mechanics of corporate layoffs—contrasting direct labor substitution against budget-shifting ("machete-style" cuts to fund AI infrastructure) and Jevons' Paradox (where demand for specialized talent rises while traditional roles are displaced). Most importantly for Private Equity deal teams, Paul outlines how to spot "AI-washing" during M&A due diligence. He explains why treating generic headcount reduction as an "AI efficiency story" is a dangerous trap: one-off fat cutting does not create a durable, recurring flow of EBITDA expansion, and overpaying by applying an expanded AI valuation multiple to one-time cost cuts destroys deal returns. Key Takeaways: - Substitution Effect vs. Budget Reallocation: Not all AI-era layoffs stem from automated productivity. Megacap and growth firms often make broad cuts across non-AI divisions simply to free up cash flow for massive AI infrastructure and token OPEX budgets. - Jevons' Paradox; Organizational Complementarity: Labor market disruption and hiring expansion coexist. Indiscriminate headcount cuts risk destroying organizational complementarity—slicing out the "glue work" and domain knowledge required to run portfolio companies effectively. - Discerning "AI-Washing" in Deal Underwriting: During M&A due diligence, deal teams must probe whether projected headcount savings stem from a proven, repeatable AI workflow or basic "fat cutting." - The Valuation Multiple Trap: Basic fat cutting has always been a standard PE value creation lever. Slapping an "AI story" onto generic cost reduction does not turn a one-off savings into a durable EBITDA engine, and boards should be extremely wary of applying higher valuation multiples under the guise of AI.

    AI-Washing Layoffs, Budget Reallocation, and the Valuation Multiple Trap
  4. Oct 21

    The State of Portfolio Ops 2026: AI Ops Titles, Forward Deployed Engineers, and the EBITDA Capture Gap

    How are elite Private Equity firms restructuring their internal Value Creation and Portfolio Operations teams to capture AI returns? In Episode 20 Paul and Dave welcome Dante Sacco, Head of Network at OneGuide, to deconstruct the findings from OneGuide’s State of Portfolio Ops Report 2026 (analyzing 60+ top PE/growth equity firms and 30+ value creation leaders). Dante reveals a dramatic shift in PE org charts: the percentage of PE firms with dedicated AI ops titles jumped from 5% to 24% year-over-year. More importantly, 83% of these hires are deeply technical—with firms deploying "Forward Deployed Engineers" (technical ICs) directly into PortCos rather than relying on generalist strategy consultants. Dante, Paul, and Dave discuss why 51% of PE firms see major functional efficiency gains in a majority of PortCos, yet only 6% are capturing 10%+ EBITDA lift—highlighting the critical gap between license adoption and true system/SDLC redesign. Key Takeaways: - The Rise of the AI Ops Title (5% to 24%): Dedicated AI roles inside PE value creation teams expanded 5x year-over-year. Tech-titled ops roles are now the single most common functional specialist position in PortCo ops teams—outnumbering both Go-To-Market and Finance specialists. - The Shift to Forward Deployed Engineers (FDEs): 83% of new AI ops hires have technical backgrounds (36% technical leaders, 45% technical individual contributors). PE firms are deploying hands-on engineers into PortCos to execute systems integration, not just write high-level strategy decks. - The EBITDA Capture Gap (51% vs. 6%): 51% of PE firms report that over half their PortCos see functional efficiency gains from AI (led by software engineering). However, less than 6% of firms have achieved a 10%+ EBITDA or revenue lift across the majority of their portfolio—proving that handing out LLM licenses without fixing underlying processes caps financial ROI. - Data Readiness is the #1 Blocker: Surveyed PE value creation leaders ranked data readiness and data quality as the #1 obstacle to AI value creation, followed by finding verified technical talent. - Standardized Playbooks & Diagnostic Screeners: Dante Sacco outlines how leading PE ops teams are scaling impact across PortCos using standardized AI modernization screeners, tech-stack rubrics, and shared vendor frameworks. Referenced Report: OneGuide - State of Portfolio Ops Report 2026: https://www.askoneguide.com/report/ OneGuide newsletter: https://www.askoneguide.com/resources/

    The State of Portfolio Ops 2026: AI Ops Titles, Forward Deployed Engineers, and the EBITDA Capture Gap
  5. Oct 15

    Agentic Software Factories: Specialist Agents, Cutting Cloud OPEX, and Protecting Production

    When generative AI accelerates code creation by 5x or 10x, writing code ceases to be the organizational bottleneck. Every other stage of the Software Development Life Cycle (SDLC) gets severely strained. In Episode 19, Paul and Dave break down OpenAI’s internal "Agentic Software Factory" model, as reported by Gergely Orosz in [The Pragmatic Engineer](https://newsletter.pragmaticengineer.com/p/openai-software-factory). Dave and Paul contrast OpenAI’s practical, systems-engineered framework against Anthropic’s aspirational guidelines. They reveal how OpenAI deploys specialized, domain-specific review agents (security, cloud infrastructure, data) to unblock code review queues. Furthermore, they analyze the financial and operational mechanics of production monitoring: while performance-tuning agents can optimize code to run workloads on 50 servers instead of 500, slashing cloud OPEX directly into EBITDA expansion, live production outage management strictly enforces human-in-the-loop oversight to prevent catastrophic revenue loss. Key Takeaways: Unblocking the Post-Code Bottleneck: Accelerating code generation without scaling code review creates massive feature-branch inventory. OpenAI solves this by orchestrating specialized review agents (Data, Security, Infrastructure) supervised by a master agent to deliver instant architectural feedback without adding headcount. Direct EBITDA Expansion via Automated FinOps: OpenAI’s performance optimization bots analyze live workloads and propose code efficiency fixes. Running the same software on 50 servers instead of 500 slashes cloud infrastructure OPEX, directly expanding portfolio company EBITDA margins. Human-in-the-Loop Production Safety (Hadfield's Law): While pre-production workflows are heavily automated, live production outage bots do *not* autonomously push fixes to live environments. Adhering to Hadfield's Law ("there's no problem so bad that you can't make it worse"), outage bots act as investigative assistants for human engineers, protecting top-line revenue and preventing customer churn. SDLC Automation in the Large: Scaling software development with AI requires treating delivery pipelines like automated server patching: moving from manual, low-volume reviews ("in the small") to continuous, agent-driven orchestration ("in the large").

    Agentic Software Factories: Specialist Agents, Cutting Cloud OPEX, and Protecting Production
  6. Oct 6

    Rethinking Incentives: (AI)ms vs. Goals and the Economics of Experimentation

    When PE backed companies attempt to drive AI R&D performance, they often set local targets—like story point velocity or code commit volume that inadvertently trigger perverse incentives and operational chaos. In Episode 18 we examine organizational incentives in the age of AI through an economic lens. Quoting W. Edwards Deming’s famous warning: "I achieved my goal, but not my aim", we break down the crucial difference between rigid tactical goals and overarching strategic aims. Paul and Dave explore how elite software companies use AI not just to write bulk code, but to dramatically accelerate product discovery experiments in partnership with product management. Drawing on frameworks from Marty Cagan (Silicon Valley Product Group), Teresa Torres, Ant Murphy, and Marc Benioff’s Salesforce V2MOM model, this episode provides listeners with a blueprint for aligning engineering output directly with hold-period EBITDA expansion. Key Takeaways: Deming’s Warning on Aims vs. Goals: "We translate aims to goals and then we do stupid things in the name of the goal." When management incentivizes arbitrary outputs (e.g., commit counts or strict deadlines), teams optimize for the metric rather than the business aim, resulting in costly fire drills, unmonetized code inventory, and burned-out staff. AI as an Experimentation Engine: The highest ROI on AI code generation comes from enabling rapid product discovery experiments. Instead of wasting months of developer capacity on unvalidated features, AI allows teams to test market hypotheses in days—provided continuous delivery (CD) pipelines are in place. Learning vs. Shipping (Ant Murphy’s Dilemma): If an engineering organization cannot ship software continuously, market learning stalls. Unblocking delivery bottlenecks shifts the competitive constraint from engineering capacity to market learning velocity. Cascading Alignment via Salesforce V2MOM: How Marc Benioff’s V2MOM framework ensures that every individual contributor—from senior architects to junior devs—understands how their daily work directly supports the CEO's top-line vision and the board's financial targets. Dynamic Incentives Across the Hold Period: Value creation aims evolve over the investment lifecycle—from foundational platform cleanup early in the hold period to margin optimization pre-exit. Management must dynamically adjust system incentives rather than enforcing static KPIs. Referenced Resources: - Ant Murphy’s Essay about Would you rather be able to learn faster than you can ship, or ship faster than you can learn? (https://www.antmurphy.me/newsletter/fix-delivery-first) - Marty Cagan / Silicon Valley Product Group (SVPG) - Teresa Torres: Continuous Discovery Habits (https://a.co/d/033dZekt)

    Rethinking Incentives: (AI)ms vs. Goals and the Economics of Experimentation
  7. Sep 30

    The New Calculus of Buy vs. Build: Claude Code, Undifferentiated Heavy Lifting, and Slashing COGS

    With generative AI tools making code generation faster and cheaper than ever, Private Equity deal teams and portfolio company executives are re-examining off-the-shelf SaaS software. Why pay six-figure SaaS subscriptions when an engineer can spin up a custom tool in days? In Episode 17, Paul and Dave re-evaluate the classic "Buy vs. Build" decision through an operational and P&L lens. Revisiting Jeff Bezos’s concept of "undifferentiated heavy lifting," Dave and Paul outline why code generation is only a fraction of the true cost of software ownership. They analyze real-world case studies—including how one portfolio company built a custom database engine that slashed COGS by 70%, while TZP Group built a custom CRM for $125/month, against the hidden operational taxes of 24/7 on-call rotations, uptime SLAs, and testing for data correctness. Key Takeaways: * The "Undifferentiated Heavy Lifting" Test: AI makes writing code fast, but it doesn't eliminate operational overhead. If a software tool doesn't create a defensible competitive moat, building it internally distracts valuable engineering capacity from core revenue-generating features. * The Hidden On-Call Reliability Tax: Building software means owning software. Maintaining 24/7 uptime for critical applications requires at least a 3-person on-call rotation to prevent developer burnout and avoid catastrophic production downtime (where 10 minutes of downtime can burn $100,000 in revenue). * Proving System Correctness: AI agents generate code quickly, but "hope is not a strategy". Systems processing financial data or high-volume transactions must prove 100% data correctness over time to prevent silent, compounding revenue leakage. * High-Leverage Custom Builds (The 70% COGS Slash): Building custom software *is* worth the investment when it directly transforms unit economics—such as replacing an off-the-shelf database to cut infrastructure COGS by 70% or replacing overpriced, flat SaaS databases with tailored internal workflow tools. * The Future of Enterprise SaaS: Dave presents his hot take on why SaaS vendors will soon offer deep, AI-driven customization to prevent clients from building homegrown alternatives.

    The New Calculus of Buy vs. Build: Claude Code, Undifferentiated Heavy Lifting, and Slashing COGS
  8. Sep 14

    Meta's Experiment: Why You Can’t Outsource System Ownership to AI

    When Private Equity boards attempt to capture AI returns, a common line item targeted is engineering headcount. In Episode 16, Paul and Dave analyze a recent LeadDev report detailing Meta’s attempt to shrink engineering teams around AI agents. While Meta saw a staggering 220% increase in code commits, only 36% made it into production, while technical incidents jumped 40% and time spent firefighting production outages exploded by 70%. Corroborated by broad industry data from DX showing a precipitous drop in Developer Experience (DevEx) scores, Dave and Paul break down why "token-maxing" without human system ownership creates massive P&L liabilities. This episode outlines why "two-pizza teams" (5 to 8 engineers) remain the optimal operational unit, why replacing developers with agents starves your system of critical operational memory, and how outage firefighting actively destroys EBITDA. Key Takeaways: - The 220% vs. 36% Gap: Meta's internal experiment proved that generating code volume with AI agents does not translate to revenue-generating features. Commits skyrocketed 220%, but actual production shipping only grew 36%. - The 70% Firefighting Tax: Replacing developer oversight with autonomous agents caused technical incidents to spike 40% and developer firefighting time to surge 70%. Every hour spent firefighting outages in production is paid labor lost to reactive crisis management rather than value creation. - System Ownership Cannot Be Outsourced: You cannot outsource the operational responsibility of running a company to AI models trained on generic public code. When complex production systems break, human engineers who understand the architecture must be available to fix them. - The Return of the Two-Pizza Team: Meta’s research concluded that the ideal team size remains 5 to 8 engineers. Shrinking teams below this threshold creates severe on-call burnout, rotation failures, and institutional risk. https://leaddev.com/ai/meta-tried-to-shrink-engineering-teams-around-ai

    Meta's Experiment: Why You Can’t Outsource System Ownership to AI

Ratings & Reviews

5
out of 5
2 Ratings

About

Engineering Alpha in Private Equity is a podcast about how software engineering and data science excellence create operational alpha. Hosted by Dave Mangot, the author of _DevOps Patterns for Private Equity_, who has worked with operating teams at Thoma Bravo, Hg Capital, and other top-tier PE firms. Co-hosted by Paul Karner, PhD, an economist with two decades inside PE-backed companies. Each episode explores the intersection of technology decisions and investment outcomes.

You Might Also Like