“[Using AI is] also a cultural question, because you need to break down silos in old ways of doing things and get the people on board with doing things in a different way.” – James Brocklebank – Co-chair of Advent International

Private equity’s embrace of artificial intelligence is not primarily a story about algorithms; it is a story about organisational behaviour, power structures, and how investment decisions are made and challenged inside firms that deploy vast pools of capital.^1,13 The tension lies between highly codified, committee-based processes built over decades and a new class of systems that can ingest years of investment history, detect patterns no human can see, and propose different ways of working.^13,16 In that space, the central obstacle is rarely technical capability. It is whether senior dealmakers, sector teams and operating partners are willing to dismantle entrenched silos and accept a more transparent, data-driven culture in which their judgement is consistently interrogated by machines.

From bespoke judgement to codified memory

Private equity decision-making has historically depended on the tacit knowledge of partners who have seen multiple cycles, negotiated complex transactions, and built mental models for what makes a good deal. That knowledge is embedded in narratives, committee debates and deal documentation rather than in formally structured datasets.^10,27 When an investor trains an AI system on more than a decade of investment committee papers, they convert that tacit institutional memory into an explicit, queryable asset that can be accessed by anyone in the organisation.^13,16 The system can surface how similar deals were debated, what risks were emphasised, where views diverged, and which decisions ultimately performed well or poorly. It changes who can participate meaningfully in discussions, because associates, principals and new partners gain direct visibility into historical reasoning that was previously accessible only through oral tradition.

That shift threatens traditional hierarchies. If the machine can show that a given pattern of argumentation has repeatedly led to underperformance in a specific subsector, it implicitly challenges the authority of those whose intuition aligns with that pattern.¹³ For firms built on the reputational capital of star dealmakers, such transparency is destabilising. The cultural question is whether leaders treat this as an opportunity to improve collective judgement or as an intrusion into their autonomy. The practical reality is that without senior sponsorship, the AI corpus becomes a curiosity used by a few enthusiastic analysts rather than a tool that genuinely influences which deals reach the term sheet stage.

Silo mentality inside the deal lifecycle

Large private equity houses tend to organise around sector teams, regional offices, and specialised functions such as deal sourcing, due diligence, financing, and portfolio value creation.^7,10,25 Each of these units develops its own language, metrics, and processes. Over time, this produces what management literature calls a silo mentality: a reluctance to share information across divisions, a tendency to optimise for local objectives rather than firm-wide outcomes, and defensive reactions when external scrutiny increases.^28,12 In an AI-enabled firm, these silos are not merely inefficient; they actively degrade model performance. Fragmented data stores, heterogeneous document standards and inconsistent tagging mean that models trained on one team’s inputs may misinterpret another team’s outputs or fail to capture crucial cross-functional context.^2,5

Consider deal sourcing and due diligence. AI engines can scan news, filings, thematic reports and proprietary databases to identify targets and assess risks at scale.^2,5,11 Yet if sourcing teams do not regularly feed back which algorithmically identified leads convert into credible opportunities, or if diligence teams do not annotate which red flags proved material post-acquisition, the training loop remains broken. The machine continues to propose deals based on historical criteria that may no longer reflect the firm’s strategic focus or risk appetite. Silos, in other words, make AI stupider. Breaking those silos requires shared data taxonomies, common documentation standards, and cross-functional governance that binds teams to a unified view of value creation rather than isolated scorecards.^12,24

Strategic tension: augmentation versus automation

Another fault line runs between two visions of AI in private equity: augmentation of human judgement versus partial automation of investment decisions. Most credible practitioners argue that AI should act as a research assistant, pattern detector and workflow optimiser rather than a replacement for the investment committee.^11,14 Systems ingest long-run performance data, macroeconomic indicators, sector dynamics, and operational metrics to highlight correlations and scenarios the team might otherwise miss.^5,11 Human partners then weigh these insights against qualitative signals: management quality, political risk, regulatory change, or strategic fit with the firm’s portfolio.

However, as models become more powerful, they can be configured to produce ranked deal lists, recommended bid ranges, or suggested capital structures based on parameterised inputs such as expected cash flows, leverage ratios, and sector volatility.^5,14 In quantitative terms, a simple discounted cash flow model would compute enterprise value using EV = \sum_{t=1}^{T} \frac{FCF_t}{(1 + r)^t}, where FCF_t represents forecast free cash flows and r the discount rate. AI systems extend this by learning from many such models, linking them to real-world outcomes across hundreds of deals, and optimising r and FCF_t paths under different scenarios. The cultural question becomes: at what point does the firm allow algorithmic recommendations to constrain or override the preferences of individual partners? If a model warns that a highly favoured deal has a statistical profile similar to past underperformers, does the committee walk away, modify the thesis, or discount the model’s view?

James Brocklebank’s experiment: codifying committee debates

Against this backdrop, the decision by a leading investor to train an AI robot on 13 years of investment committee papers is strategically significant.^13,16 It suggests a willingness to treat the firm’s internal deliberations as data, not merely as private conversations. Each memo, debate and decision becomes an input to a system that can map how the organisation has responded to different macro conditions, sector narratives and management teams. Advent International’s scale and complexity make such an experiment particularly revealing: a global portfolio, multi-sector coverage, and cross-border teams generate a rich dataset of decision patterns that the system can mine for latent structure.^7,10

Reports indicate that the robot surfaces connections and anomalies that humans do not readily see, prompting both excitement and discomfort among users.^13,16 It may highlight that certain risk factors were consistently downplayed in bull markets, or that particular deal archetypes delivered outsize returns when combined with specific operational playbooks. For a managing partner, this is both an opportunity to refine strategy and a mirror held up to the firm’s collective biases. The investor involved, with responsibilities spanning Europe and global governance roles, is well placed to use these findings to challenge siloed behaviours and push for more horizontally integrated decision-making.^7,19,25 But the success of such a project ultimately depends on whether colleagues accept that their past reasoning is subject to machine-led scrutiny and are prepared to adjust their habits accordingly.

Breaking down silos: governance, data and incentives

Organisations seeking to replicate such an AI-enabled investment memory cannot rely on technology alone. They need governance structures that compel siloed teams to contribute data, align on taxonomies, and participate in common forums where AI insights are debated.^11,24 Strategy and change management literature converges on several practical mechanisms. First, a unified vision and shared goals that explicitly reference AI-enabled outcomes – for example, improving hit rates on approved deals, reducing diligence cycle times, or enhancing portfolio value-creation interventions – should be communicated from the top.^18,26 Second, cross-functional teams, such as an AI steering committee comprising representatives from deal, operations, risk, IT and compliance, must own both deployment and ongoing refinement.²⁴

Third, incentives must reward collaboration. If year-end bonuses and promotion criteria continue to focus on individual deal origination or sector P&L, teams will hoard information and treat AI tools as personal advantages rather than shared infrastructure.^24,18 Aligning compensation with firm-wide metrics – portfolio-level EBITDA improvement, realised IRR over a multi-year horizon, or success rates of AI-informed value-creation playbooks – pushes behaviour towards openness. From a more technical perspective, firms also need to federate data across silos while respecting regulatory and privacy constraints. That typically means building data lakes or mesh architectures where investment memos, financial models, operational KPIs and market intelligence are tagged and accessible through governed interfaces, enabling models to draw connections while audit trails preserve accountability.^11,23

Cultural resistance: fear of exposure and loss of control

Resistance to such transformations does not stem solely from legacy workflows. It often arises from fear of exposure and loss of control. If AI can compare one team’s performance against another’s, or correlate specific decision styles with outcomes, the relative performance of individuals becomes starkly visible.^12,28 Senior investors who have built their reputation on anecdotal wins may be reluctant to see their track record reframed in terms of long-run, risk-adjusted returns. There is also a legitimate concern that overly mechanistic metrics could undervalue qualitative contributions such as relationship-building or regulatory navigation.

Moreover, some practitioners worry that collapsing silos entirely may overload teams with information and meetings, replacing focused execution with constant cross-functional coordination.¹⁵ Change management experience suggests that the goal should be not to destroy all boundaries but to create purposeful connections: shared objectives where they matter, structured hand-offs between teams, and regular forums for debate around AI outputs.^15,24 Well-designed AI tooling can help by synthesising complex inputs into concise dashboards, highlighting only the most salient risks, anomalies and opportunities. Cultural adaptation then becomes a question of training, leadership modelling, and gradual exposure rather than abrupt reorganisation.

Debates and objections: can private equity stay differentiated?

There is an active debate about whether widespread adoption of AI will commoditise private equity. If every large firm deploys similar models trained on overlapping external data – filings, market feeds, macro series – will they all converge on the same deals and strategies?^5,14 One counterargument is that differentiation lies in proprietary data and the ability to integrate operational insights from portfolio companies into investment decision-making. AI systems that ingest on-the-ground performance metrics, customer churn patterns, pricing experiments and supply chain disruptions across hundreds of assets can generate unique signals about how specific business models respond to shocks.^5,20 Firms that collapse silos between investment teams and operating partners, and that codify value-creation playbooks into the AI stack, may gain an edge.

Another objection centres on model risk. AI recommendations are only as robust as the data and assumptions embedded within them.^11,23 Historical investment committee papers capture the firm’s past biases as well as its wisdom. If the organisation has systematically avoided certain regions, technologies or founder profiles, the robot may infer that such deals are unattractive even if the external world has changed. That makes human oversight and explicit challenge processes essential. Governance frameworks emphasise accountability, bias mitigation, explainability, and the ability to override automated outputs when they conflict with strategic priorities or ethical considerations.¹¹ Breaking down silos helps here too: diverse teams reviewing AI outputs are more likely to spot blind spots than a single homogeneous group.

Why the cultural question matters now

Several trends make this cultural dimension urgent. First, private equity has invested more than USD 1 trillion in information technology since 2020, much of it aimed at digital and AI-enabled capabilities across portfolios.¹⁷ Limited partners increasingly expect general partners not only to back AI-native businesses but also to deploy AI in their own underwriting and monitoring processes. Second, regulators and societal stakeholders are scrutinising algorithmic decision-making for fairness, transparency and systemic risk, particularly when large capital allocators are involved.^11,23 Firms that cannot demonstrate coherent governance and cross-functional alignment around AI may face scepticism, both commercially and in policy debates.

Third, competition is intensifying. Consulting analyses suggest that only a minority of private equity firms can show meaningful, generalisable returns from AI across their portfolios.^14,8 Those that have done so typically invest in centralised AI operating models with defined components: governance, strategy alignment, data federation, model evaluation, architecture optimisation and operating playbooks.⁸ Each of these components presupposes cultural willingness to collaborate beyond traditional boundaries. Without that, AI experiments remain tactical – a sourcing tool here, a document parser there – rather than re-shaping how the firm perceives and manages risk.

Implications for leadership and organisational design

For leaders in positions similar to James Brocklebank’s – co-heading geographic franchises, sitting on global executive committees, and guiding investment strategy – the challenge is to turn AI-enabled insight into durable organisational change.^7,19,25 That means moving beyond pilot enthusiasm to institutionalisation. Practical steps include mandating that all new investment committee papers conform to structured templates compatible with the AI corpus, embedding AI-generated analysis sections into standard memo formats, and allocating time in committee agendas specifically to discuss the machine’s perspective.^13,16 Over time, this normalises the presence of AI in high-stakes discussions rather than treating it as an optional add-on.

Organisational design may evolve accordingly. Firms might appoint AI champions within each sector team responsible for maintaining data pipelines, collecting feedback, and liaising with central data science units. Training programmes would focus not only on how to use the tools but on how to interpret their limitations, including understanding that correlation does not equal causation and that statistical confidence intervals do not absolve decision-makers of responsibility. Some may experiment with rotational programmes where investors spend time in data and technology teams, reducing mutual misunderstanding between dealmakers and engineers. All these moves aim to erode silo walls by creating shared language and joint ownership of AI outcomes.

Looking ahead: AI as a permanent participant in investment debates

If these cultural and structural shifts succeed, AI will become a permanent participant in investment debates – not an oracle, but a disciplined voice that consistently surfaces historical patterns, alternative scenarios and previously overlooked connections. The presence of such a voice changes how disagreements are framed. Rather than arguing purely from personal experience, partners will increasingly reference model outputs, stress tests and cross-portfolio comparables. That does not eliminate politics or judgement; it channels them through a more transparent evidentiary layer.

The broader backstory to the statement about culture and silos, then, is the emergence of private equity firms as data-intensive institutions whose competitive edge depends as much on how they organise information and people as on how they source and price deals.^10,17 In that world, the hardest challenge is persuading seasoned professionals to accept that doing things in a different way – exposing their decisions to machine-led scrutiny, sharing data across boundaries, co-designing AI-driven processes – is not a threat to their craft but a route to making that craft more resilient. The outcome of this cultural negotiation will determine which firms simply experiment with AI and which rebuild their investment engines around it.

 

References

1. ‘Excited and terrified’: One of private equity’s top investors built an AI … – 2026-05-26 – https://fortune.com/2026/05/26/james-brocklebank-advent-international-goldman-sachs-ic-robot/

2. James Brocklebank – Managing Partner at Advent – LinkedIn – https://www.linkedin.com/in/jamesbrocklebank

3. AI in Private Equity: Top Use Cases You Need To Know – SmartDev – 2025-09-05 – https://smartdev.com/ai-use-cases-in-private-equity/

4. How to break down silos in your organization: tips and best practices – 2021-12-01 – https://triskellsoftware.com/blog/how-to-break-down-silos/

5. ‘Complexity is our friend’: James Brocklebank on private equity, AI … – 2026-06-02 – https://www.adventinternational.com/ideas/complexity-is-our-friend-james-brocklebank-on-private-equity-ai-and-the-advent-edge/

6. AI in Private Equity 2026: Use Cases and Data | Blott – 2026-04-01 – https://www.blott.com/reports/ai-use-cases-in-private-equity

7. Breaking Down Communication Silos: 4 Essential Practices for … – 2024-08-29 – https://blog.carnegieinvest.com/nonprofit-investment-services-blog/breaking-down-communication-silos-4-essential-practices

8. James Brocklebank – Advent International – 2026-04-20 – https://www.adventinternational.com/our-team/james-brocklebank/

9. The Portfolio Problem: Why Private Equity Can’t AI Its Way Out One … – 2026-03-05 – https://www.linkedin.com/pulse/portfolio-problem-why-private-equity-cant-ai-its-way-out-smith-pnaxe

10. 5 Leadership Development Tactics to End Silos – 2023-02-28 – https://www.virginiamasoninstitute.org/blog/5-tactics-to-break-down-silos-and-support-cross-functional-collaboration

11. ‘Complexity Is Our Friend’: James Brocklebank on Advent’s Private … – 2026-05-26 – https://www.goldmansachs.com/insights/goldman-sachs-exchanges/complexity-is-our-friend-james-brocklebank-on-advents-private-equity-strategy

12. AI Governance in Private Equity and Asset Management [2026] – Atlan – 2025-09-26 – https://atlan.com/know/ai-governance-private-equity-asset-management/

13. Breaking Down Silos in the Workplace: A Framework to Foster … – 2024-09-20 – https://pmc.ncbi.nlm.nih.gov/articles/PMC11419935/

14. Inside the AI-First Private Equity Firm | BCG – 2026-01-23 – https://www.bcg.com/publications/2026/inside-the-ai-first-private-equity-firm

15. Breaking down silos during an org change. What’s actually worked … – 2025-08-15 – https://www.reddit.com/r/changemanagement/comments/1mqyfri/breaking_down_silos_during_an_org_change_whats/

16. ‘Excited and terrified’: One of private equity’s top investors built an AI … – 2026-05-26 – https://finance.yahoo.com/sectors/technology/articles/excited-terrified-one-private-equity-120000499.html

17. Private Equity Propels America to the Front of the AI Race – 2025-03-10 – https://www.investmentcouncil.org/private-equity-propels-america-to-the-front-of-the-ai-race/

18. Breaking Down Organizational Silos for Better Collaboration – Chronus – 2026-06-26 – https://chronus.com/blog/organizational-silo-busting

19. James Brocklebank – Alternative Investments Conference – https://www.lseaic.com/speakers-2024/james-brocklebank

20. AI is a double-edged sword for PE portfolios – 2026-02-12 – https://www.privateequityinternational.com/ai-is-a-double-edged-sword-for-pe-portfolios/

21. Breaking Down Silos: How Integrations Are Revolutionizing Project … – 2025-09-03 – https://arcoro.com/resources/breaking-down-silos-how-integrations-are-revolutionizing-project-workflows

22. Goldman Sachs Leverages Local Expertise for European Advantage – 2026-06-02 – https://www.linkedin.com/posts/advent-international_james-brocklebank-on-private-equity-ai-activity-7467592692462403584-6Akb

23. AI adoption in private equity: insights and challenges – Pictet – 2025-07-01 – https://www.pictet.com/us/en/insights/megatrending-2025-investment-opportunities/private-equity-and-ai

24. [PDF] Seven strategies for breaking down silos Dealing with market … – https://www.strategyand.pwc.com/gx/en/insights/2016/dealing-market-disruption/dealing-with-market-disruption.pdf

25. James Brocklebank – Level20 – 2022-06-13 – https://www.level20.org/team/james-brocklebank/

26. 3 Tips for Breaking Down Silos and Fostering Collaboration – 2018-11-28 – https://www.cooperative.com/conferences-education/overview/Pages/Traceys-Takeaways-3-Tips-for-Breaking-Down-Silos-and-Fostering-Collaboration.aspx

27. James Brocklebank on the Shifting Tectonic Plates of Private Equity – 2025-01-07 – https://www.adventinternational.com/news/james-brocklebank-on-the-shifting-tectonic-plates-of-private-equity/

28. Understanding Silo Mentality in Business: Causes and Solutions – https://www.investopedia.com/terms/s/silo-mentality.asp