hub
PRISMVEST
New York Stock Exchange
The Final Frontier

From Averages to Intelligence

PrismVest is the Next Operating System for Investment Management — building scalable digital cognitive capacity, proprietary institutional memory, and human-directed agent networks.

Explore Architecture
Cognitive Scaling Theorem
PrismVest Capability = Compute × Algorithmic Efficiency × Agentic Improvements

The fundamental scaling law of agentic investment systems: multiplying raw compute and quantitative algorithms through human-governed multi-agent cognitive capacity.

Strategic Perspective: Scalable Digital Labor
arrow_forward
psychology

From Software Tools to Digital Cognitive Capacity

Traditional investment technology requires humans to operate software tools at every step. PrismVest introduces a new architecture: specialized AI agents perform continuous data development, earnings revisions, factor research, regime analysis, and tax optimization under explicit human oversight.

check_circle Multi-Agent Coordination
check_circle Human Fiduciary Conductor
account_balance

Proprietary Memory

Reconstructing historical investment decisions into a structured learning ledger: Information → Forecast → Decision → Outcome → Learning.

trending_up

The Conductor Archetype

The portfolio manager acts as orchestrator—establishing objectives, risk budgets, and permissions while autonomous agent networks execute analytical tasks.

Five Pillars of Cognitive Operating Leverage

Pillar 1

Research Leverage

Dramatically expanding the scope of corporate filings, transcripts, and alternative datasets analyzed without linear headcount growth.

Pillar 2

Decision Leverage

Giving portfolio managers simultaneous access to multiple specialized signals: regimes, quality metrics, earnings revisions, and tax lots.

Pillar 3

Customization

Automating gain budgets, client restrictions, and factor tilts across hundreds of separately managed accounts (SMAs).

Pillar 4

Closed-Loop Learning

Supervised systematic error attribution connecting forecasts to realized outcomes to continuously improve institutional models.

Systems Architecture

Portfolio Management & Architecture

Re-engineering workflows from manual terminal search tools to coordinated digital labor.

Active Agent Domains
Specialized Domains

Data, Regimes, Factors, Changing-Economy, Tax, & Strategy Optimization.

Institutional Decision Ledger
10+ Year History

Reconstructed point-in-time forecasts, rationales, and realized return outcomes.

Analytical Capacity Growth vs. Traditional Headcount Scaling
Operating Leverage Multiplier
Data Curation
Earnings Scans
Regime Tests
Tax Harvesting
Optimization
Active Cognitive Workflow Layers
Supervised Execution
Data Validation & Ingestion

Ensuring point-in-time integrity, missing-data handling, and outlier controls.

Active
Factor & Regime Deliberation

Cross-evaluating momentum, quality, revisions, and macro states.

Active
Fiduciary Exception Verification

Challenging model assumptions and generating human approval packets.

Continuous
Academic / Institutional White Paper

The Academics of Intelligence

August 2026 | Jon DuPrau / PrismVest | Institutional Research
open_in_new

FROM AVERAGES TO INTELLIGENCE

PrismVest is the Next Operating System for Investment Management

Academic / Institutional White Paper • August 2026

Abstract

PrismVest represents the next stage in a 140-year evolution of investment technology: from market averages, to capitalization-weighted benchmarks, to investable indexes, to enhanced and direct indexing, and now to human-supervised agentic investment systems. The central thesis is that investment management will increasingly be organized around networks of specialized AI agents that research, test, coordinate, remember, and recommend under explicit human-defined objectives, permissions, risk controls, and fiduciary governance. The portfolio manager does not disappear; the role moves upward from executing every analytical step toward orchestrating intelligence, allocating risk, resolving ambiguity, and retaining accountability.

Recent research makes this direction increasingly credible while counseling restraint. Machine-learning methods have demonstrated economically meaningful predictive gains in asset-pricing research; controlled field experiments show generative AI can raise knowledge-worker productivity; and new 2025–2026 research explicitly studies multi-agent systems for economic research and portfolio management. CFA Institute emphasizes human judgment, explainability, model governance, and accountability, while SEC enforcement against “AI washing” demonstrates that claims about investment use of AI must be precise and supportable. PrismVest should therefore be understood as a forecast and operating model—not a promise of autonomous outperformance.

Executive Summary

Indexing began as information compression. Dow’s average reduced a complex market to a reproducible signal; capitalization weighting later made the signal more economically representative.

The benchmark became a portfolio operating system. Index funds, ETFs, factors, enhanced indexing, custom indexing, and direct indexing progressively separated universe, signal, weighting, risk, tax, and implementation decisions.

PrismVest extends this logic from governed algorithms to governed intelligence. Specialized AI agents perform narrow investment functions while portfolio managers define objectives, constraints, permissions, and accountability.

AI can reduce the marginal cost of cognitive work, allowing investment organizations to increase analytical coverage and output without proportionate increases in headcount.

Proprietary decision history can become a strategic asset. A structured record of forecasts, rationales, actions, outcomes, and errors can create a firm-specific learning environment that general-purpose models do not possess.

As generalized AI becomes more available, differentiation should migrate toward proprietary data, institutional context, workflow design, evaluation systems, governance, client trust, and coordination.

PrismVest’s business model is a productivity and intellectual-capital model: broader research coverage, scalable customization, stronger documentation, and potentially higher operating leverage, subject to rigorous validation and fiduciary controls.

The Long Arc: From Averages to Intelligence

The intellectual foundation for PrismVest is strongest when presented not as a break with indexing, but as the next extension of a long progression. Average Becomes the Market begins with a simple observation: before a market average existed, investors could observe individual securities but could not easily answer, “How is the market doing?” Dow’s innovation was a disciplined summary. Standard & Poor’s improved the measuring stick, mutual funds adopted indexes as report cards, and passive funds eventually allowed investors to own the report card itself.

From Dogs to Drivers of the Dow shows how a benchmark can become a parent universe rather than a final portfolio. Dogs represented an early democratized systematic rule: one visible variable, dividend yield, selected a subset of the Dow. Drivers represented a broader proposition—use multiple forms of evidence, including leadership, fundamentals, earnings information, quality, and risk controls, rather than assume every laggard will mean revert. The historical transition was from a static benchmark to systematic enhancement.

From Averages to Algorithms generalizes the point. The defining feature of modern indexing is not passivity; it is governance by an explicit methodology. Once computing, data, optimization, fractional shares, and low trading costs became available, investors could separate decisions that earlier indexes bundled together: universe, information set, weighting, risk budget, tax management, and implementation.

Market Average → Benchmark → Investable Index → Enhanced Algorithm → Agentic Investment System.

2. Why the Next Step Is Agentic Rather Than Merely Algorithmic

Traditional quantitative systems are powerful but largely procedural. They ingest specified data, calculate specified transformations, apply specified rules, and return outputs. Machine learning expanded that architecture by allowing relationships to be inferred from data rather than fully hand-coded. Gu, Kelly, and Xiu’s Empirical Asset Pricing via Machine Learning demonstrated that nonlinear methods can capture interactions missed by conventional specifications and improve out-of-sample risk-premium forecasts. The result does not prove that machine learning always outperforms; it establishes that richer computational methods can extract economically relevant structure from high-dimensional financial data.

Foundation models add natural-language reasoning across unstructured information. Filings, transcripts, news, research papers, investment-policy documents, compliance manuals, portfolio commentary, and historical decision notes can increasingly be analyzed within the same computational environment. CFA Institute’s 2025 AI in Asset Management volume describes AI and machine learning as reshaping portfolio design, risk oversight, trading, research, and client engagement, while stressing human judgment and ethical governance.

Agentic AI adds a further step. An agent can be given an objective, plan intermediate steps, call tools, retrieve information, evaluate results, retain context, and continue until a bounded task is completed. Anton Korinek’s 2025 NBER paper demonstrates agents that conduct literature reviews, write and debug econometric code, retrieve economic data, and coordinate multi-step research. Hadfield and Koh examine the broader possibility that such systems will plan and execute complex tasks with progressively less direct oversight.

Software calculates. Algorithms rank. Agents increasingly coordinate work.

3. The PrismVest Architecture: Digital Labor for Investment Management

The PrismVest Thesis describes a hybrid organization combining experienced portfolio managers with specialized AI agents. That framing is increasingly consistent with emerging academic work. Caner, Capponi, Sun, and Tan (2026) propose an agentic portfolio platform in which specialized LLM agents separately analyze fundamentals and news sentiment and deliberate over candidate securities. Ang, Azimbayev, and Kim’s 2026 Self-Driving Portfolio describes roughly 50 specialized agents generating capital-market assumptions, constructing portfolios using multiple methods, critiquing one another, and comparing prior forecasts with realized outcomes, while an Investment Policy Statement constrains the system. These papers are early and are not settled evidence of superior live performance, but they independently validate the architectural direction envisioned by PrismVest.

PrismVest can organize specialized digital labor around six domains: Data Development and Management; Regime Detection; Factor and Signal Analysis; Changing-Economy Research; Tax and Implementation Optimization; and Strategy Weighting and Portfolio Optimization. The orchestration layer sits above them. The portfolio manager defines the strategy objective, eligible universe, risk budget, investment horizon, permitted tools, escalation rules, and implementation authority. Agents perform bounded tasks. Verification agents challenge evidence, identify missing data, test sensitivity, and flag conflicts. Portfolio construction converts validated signals into proposed positions. Human governance remains responsible for approval, fiduciary judgment, methodology changes, and exceptional conditions.

Strategy Objective → Human Governance → Specialized Agents → Research and Testing → Portfolio Recommendation → Risk/Compliance Verification → Human Authorization → Implementation → Outcome Measurement → Learning

4. Proprietary Data Becomes Institutional Memory

General-purpose models are likely to become broadly available. If competitors can access similar foundation models and computing infrastructure, the differentiated asset becomes the information and learning environment surrounding those models.

The proposed decision ledger is therefore more than a database. PrismVest can reconstruct historical stock-selection and portfolio decisions, beginning with the prior decade where records permit, and capture new decisions with much greater structure. Each observation can include the information set available at the time, forecast, factor scores, qualitative rationale, regime assumptions, portfolio action, risk constraints, expected horizon, subsequent return, benchmark-relative result, realized fundamental developments, and whether the original thesis was confirmed or falsified.

That architecture turns investment history into supervised institutional memory. The critical distinction is between memorization and generalization. A system that perfectly explains past decisions but fails on unseen data has not learned a useful investment relationship. Research therefore must use point-in-time data, holdout samples, walk-forward testing, stability analysis, transaction costs, and explicit controls against look-ahead and survivorship bias.

The strategic asset is not simply more data. It is structured feedback connecting decisions to outcomes.

5. From Human Productivity to Organizational Productivity

The economic case for PrismVest begins with the declining marginal cost of cognitive work. Brynjolfsson, Li, and Raymond found that a generative-AI assistant increased customer-support productivity by 14% on average and 34% for novice and lower-skilled workers. Dillon, Jaffe, Immorlica, and Stanton’s 2025 randomized field experiment across 66 firms and 7,137 knowledge workers found meaningful reductions in time spent on email among active AI users, although individual access alone produced less evidence of broad organizational redesign.

That distinction is central to the PrismVest forecast. A copilot can make an analyst faster. An agentic operating model can change how the firm divides and coordinates work. Dell’Acqua and coauthors’ 2025 field experiment at Procter & Gamble found that individuals working with AI could match the performance of conventional two-person teams on product-innovation tasks. The implication is not that AI universally replaces teams; it is that the production function of knowledge work can change when intelligence becomes inexpensive and rapidly deployable.

Investment firms historically scaled cognitive output primarily by adding analysts, portfolio managers, operations staff, and technology seats. PrismVest proposes a different scaling function: senior human judgment + proprietary data + reusable agent workflows + computing. If the system works as intended, research coverage, testing frequency, documentation, tax analysis, portfolio scenario analysis, and client-specific customization can expand faster than payroll.

6. The Portfolio Manager Becomes an Orchestrator

The PrismVest thesis does not require the disappearance of the portfolio manager. It predicts a change in comparative advantage. When data gathering, document reading, first-pass modeling, scenario generation, and routine monitoring become cheaper, scarce capabilities move toward objective setting, judgment under uncertainty, strategy design, risk allocation, client communication, governance, and accountability.

This is consistent with the Financial Analysts Journal’s 2025 discussion of the “disappearing edge” of the discretionary portfolio manager, which argues that the role may evolve from sole decision-maker toward allocator and steward of machine-driven models. CFA Institute likewise frames AI as supplementing professional judgment rather than eliminating it.

Human employees + software tools → Human employees + AI copilots → Humans supervising specialized agents → Networks of humans and governed autonomous agents

At the mature stage, portfolio managers need not personally instruct every agent. They define the objective function, resources, permissions, risk limits, escalation conditions, and evidence standards. Specialized agent teams can work continuously on data quality, earnings revisions, factor efficacy, economic regimes, tax lots, competitive advantage, and portfolio constraints. The human manager remains responsible for deciding what the system is trying to accomplish and whether its proposed action is appropriate.

7. From Information Scarcity to Coordination Scarcity

Charles Dow’s competitive advantage was access to timely information. In the modern market, real-time prices, filings, transcripts, estimates, news, and economic statistics are broadly distributed. The earlier papers describe how information abundance raised the hurdle for active management. PrismVest extends that observation: the emerging constraint is not merely information access but coordination.

Investment decisions combine datasets with different identifiers, frequencies, definitions, and error structures. Fundamental research may conflict with price trends; earnings revisions may conflict with valuation; tax objectives may conflict with desired active weights; client restrictions may conflict with model rankings. Much of professional portfolio management consists of reconciling these competing claims under time pressure.

Agentic systems can lower coordination costs by creating a common decision architecture. Specialized agents can produce standardized evidence packets, communicate uncertainty, route exceptions, and preserve an audit trail. A portfolio manager can supervise interaction among analytical disciplines rather than manually assemble every input.

AI → Better Coordination → New Workflows → Organizational Restructuring → New Business Models → Shifts in Competitive Advantage

8. The Business Model: From Software Seats to Cognitive Capacity

Traditional investment technology is usually purchased as software: terminals, data feeds, research databases, risk systems, accounting systems, and portfolio-management platforms. Their economic value is largely measured by how much more productive they make employees. Agentic systems blur that boundary because software begins performing portions of the work itself.

The relevant cost comparison therefore moves from software price per seat toward cost per completed analytical workflow, decision supported, portfolio monitored, tax lot optimized, or client account customized. This creates the possibility that the addressable market for investment technology migrates from the technology budget toward portions of the labor budget.

For PrismVest, the business model can be expressed through five reinforcing sources of operating leverage: research leverage, strategy leverage, customization leverage, governance leverage, and learning leverage. Research leverage expands securities and datasets reviewed without proportional analyst growth. Strategy leverage reuses modules across universes, factors, and themes. Customization leverage systematizes tax budgets, restrictions, and client preferences. Governance leverage standardizes evidence, provenance, validation, and exception handling. Learning leverage makes each completed decision part of the proprietary decision history.

The economic objective is not “replace people.” It is to increase the amount of investment intelligence the firm can coordinate per dollar of operating expense while preserving fiduciary responsibility.

9. PrismVest as the Next Stage of Direct and Custom Indexing

The progression from algorithms to agents is especially natural in direct indexing. A pooled index product must apply one portfolio to many investors. A direct-index account can already incorporate individual securities, tax lots, exclusions, legacy holdings, gain budgets, factor tilts, and tracking constraints. Academic and practitioner research on direct indexing demonstrates its ability to support tax management and investor-specific customization, while emphasizing that benefits depend on the investor’s circumstances.

Agentic architecture can make customization more scalable. A tax agent can evaluate harvesting opportunities; a restriction agent can verify prohibited securities; a transition agent can compare realized-gain budgets; a factor agent can preserve desired exposures; a risk agent can estimate tracking error; and a portfolio-construction agent can reconcile competing objectives. The system can generate a proposed trade list together with an explanation of which constraints drove each deviation.

Dow compressed many securities into one number. Direct indexing used computing to reconstruct the index at the security level. PrismVest adds another layer: the reconstructed portfolio can become dynamically personalized by a governed network of analytical agents.

10. Competitive Advantage in a World of Abundant Intelligence

The PrismVest thesis contains an important paradox. If artificial intelligence becomes widely available, intelligence itself may become less scarce. Economic value should then migrate toward complementary resources that remain difficult to reproduce.

For an investment firm, those scarce complements may include proprietary decision history, clean point-in-time datasets, tested workflows, institutional knowledge, client relationships, trusted distribution, regulatory permissions, risk infrastructure, human judgment, and organizational reputation. The winning firm may not be the one with exclusive access to the most powerful general model. It may be the firm that most effectively combines general intelligence with proprietary context and disciplined implementation.

This mirrors the bottleneck logic developed in Powering Compute & Intelligence. When one constraint is relieved, value migrates toward the next constraint. In investment management, generalized model capability may become abundant while high-quality proprietary context, judgment, governance, and coordination remain scarce.

General AI + Proprietary Context + Workflow Architecture + Governance + Distribution + Trust = Coordination Advantage

11. Governance Is the Product, Not an Appendix

As autonomy increases, governance becomes more important, not less. CFA Institute’s recent work emphasizes transparency, accountability, explainability, and human oversight. SEC enforcement actions against advisers that made misleading statements about AI use provide an equally important commercial lesson: firms must be able to demonstrate what the system actually does and must not market capabilities that are not implemented and controlled.

A PrismVest governance framework should include point-in-time data controls; model and prompt versioning; source provenance; permission boundaries; independent verification agents; human approval thresholds; restricted access to trading and client data; model-risk testing; cybersecurity and privacy controls; audit logs; incident escalation; reproducible research; and documented methodology-change procedures.

The most important design principle is bounded autonomy. Research agents can be highly autonomous inside a research sandbox. Portfolio agents can generate recommendations. Trade execution should require explicit permissions and controls appropriate to the strategy and regulatory environment. High-consequence changes to methodology, risk budgets, client restrictions, or portfolio objectives should remain human-governed.

12. Risks and Counterarguments

A credible forecast must acknowledge that the agentic future is not guaranteed to arrive smoothly. Large language models can hallucinate, misread context, amplify erroneous inputs, and display unstable behavior across model versions. Multi-agent systems can create the appearance of independent confirmation even when agents share the same underlying model and biases. Financial markets are low-signal, nonstationary environments in which relationships decay as participants adapt.

Model convergence presents another risk. If many firms use similar models, data, prompts, and optimization frameworks, their trades could become correlated. AI may reduce some idiosyncratic research costs while increasing crowding and systemic sensitivity to shared errors. Explainability also becomes harder as systems incorporate multiple agents, tools, memories, and nonlinear models.

There is a deeper economic counterargument: AI may commoditize analysis without producing durable alpha. Better tools can improve every participant simultaneously. Gross forecasting improvements may be competed away, leaving the primary benefits in cost, scale, tax management, client service, and operational efficiency rather than excess return. PrismVest should be designed to win under that outcome as well.

13. A Practical Development Roadmap

Phase I:

Decision Memory. Digitize and normalize historical portfolio decisions; establish a common security master; capture point-in-time inputs, rationales, forecasts, and outcomes; and build reproducible evaluation datasets.

Phase II:

Specialized Research Agents. Deploy bounded agents for literature review, filings, earnings revisions, data validation, factor research, regime analysis, tax lots, and portfolio diagnostics. Require citations, provenance, and standardized outputs.

Phase III:

Verification and Multi-Agent Deliberation. Introduce independent challenge agents, model comparison, confidence scoring, exception routing, and adversarial review. Measure whether additional agents improve decisions or merely create complexity.

Phase IV:

Portfolio Recommendation. Allow agents to generate proposed rankings, weights, transitions, and rebalance trades subject to formal risk, tax, liquidity, concentration, and client constraints. Keep human authorization as the control point.

Phase V:

Closed-Loop Learning. Compare forecasts and recommendations with realized outcomes. Attribute error to data, signal, regime, model, portfolio construction, or implementation. Update research priorities through governed methodology review rather than automatic performance chasing.

Phase VI:

Scalable Productization. Extend the architecture across enhanced indexes, thematic strategies, direct indexing, tax-aware portfolios, OCIO workflows, and customized mandates while maintaining common governance and audit standards.

14. Forecast: The Investment Firm as a Human–Machine Network

The likely destination is not a fully autonomous asset manager operating without human responsibility. It is a human–machine network in which intelligence is distributed differently. Machines will increasingly perform the work that is computationally intensive, repetitive, document-heavy, cross-sectional, or continuously monitored. Humans will increasingly concentrate on objectives, context, exceptions, client trust, strategic judgment, and accountability.

This can change the economics of scale in investment management. Historically, a boutique firm’s research capacity was constrained by headcount and access to expensive institutional systems. Agentic architecture may allow smaller organizations to deploy institutional-grade analytical coverage with fewer people, provided they possess the data, governance, investment discipline, and client relationships required to use the technology responsibly.

The forecast is therefore larger than “AI will help portfolio managers.” PrismVest proposes that investment organizations themselves will be redesigned around abundant machine intelligence. The firm becomes an orchestrator of specialized cognitive resources. Its operating leverage comes from reusable workflows; its moat comes from proprietary context and trust; its governance becomes a core product attribute; and its learning system compounds institutional knowledge over time.

Conclusion: From the Scoreboard to the Learning Organization

The story began with a scoreboard. Charles Dow compressed a market into an average so investors could understand it. Standard & Poor’s made the average more economically representative. Academic finance turned the market portfolio into a benchmark. Bogle and the index-fund industry turned the benchmark into a product. Factor, enhanced, custom, and direct indexing turned the product into a programmable portfolio architecture. PrismVest illustrates the transition from static rules toward a governed multi-agent cognitive architecture.

PrismVest is the proposed next step: the algorithm becomes a participant in the investment process.

The grand thesis is not that machines possess infallible investment judgment. It is that the cost, speed, breadth, and coordination of cognitive work are changing. Investment firms that treat AI merely as another software application may capture incremental productivity. Firms that redesign their decision architecture around proprietary data, specialized agents, continuous evaluation, and human governance may capture something more consequential: a new production model for investment intelligence.

The most durable version of PrismVest is therefore neither passive nor conventionally active, neither human-only nor autonomous. It is a governed learning organization.

Average → Benchmark → Portfolio → Algorithm → Agent → Learning System

That progression brings the history together, and provides a plausible business model for the future of investment management.

Selected Academic and Institutional References

Ang, A., Azimbayev, N., & Kim, A. (2026). The Self-Driving Portfolio: Agentic Architecture for Institutional Asset Management. SSRN.

Brynjolfsson, E., Li, D., & Raymond, L. R. (2023). Generative AI at Work. NBER Working Paper 31161.

Caner, M., Capponi, A., Sun, N., & Tan, J. (2026). Designing Agentic AI-Based Screening for Portfolio Investment. SSRN.

CFA Institute Research Foundation. (2025). AI in Asset Management: Tools, Applications, and Frontiers.

Dell’Acqua, F., et al. (2025). The Cybernetic Teammate: A Field Experiment on Generative AI Reshaping Teamwork and Expertise. NBER Working Paper 33641.

Dillon, E. W., Jaffe, S., Immorlica, N., & Stanton, C. T. (2025). Shifting Work Patterns with Generative AI. NBER Working Paper 33795.

Fabozzi, F. J., Chin, A., Yelnik, I., & Liew, J. K.-S. (2025). The Disappearing Edge: AI, Machine Learning, and the Future of the Discretionary Portfolio Manager. Financial Analysts Journal, 81(4).

Gu, S., Kelly, B., & Xiu, D. (2020). Empirical Asset Pricing via Machine Learning. Review of Financial Studies, 33(5), 2223–2273.

Hadfield, G. K., & Koh, A. (2025). An Economy of AI Agents. In The Economics of Transformative AI. NBER / University of Chicago Press.

Korinek, A. (2025). AI Agents for Economic Research. NBER Working Paper 34202.

Saha, P., Lyu, J., Saxena, A., Zhao, T., & Mehta, D. (2025). Large Language Model Agents for Investment Management: Foundations, Benchmarks, and Research Frontiers. SSRN.

Sosner, N., Gromis, M., & Krasner, S. (2022). The Tax Benefits of Direct Indexing. Journal of Beta Investment Strategies.

U.S. Securities and Exchange Commission. (2024). AI Washing statements and enforcement actions concerning investment advisers.

Source Note

This paper synthesizes the PrismVest research papers developed for this architecture with external academic and institutional research. Claims about PrismVest’s future architecture and economics are forecasts and strategic hypotheses, not established facts or guarantees of investment performance. Emerging 2025–2026 SSRN work on agentic portfolio management is prepublication research and should be evaluated accordingly.

Related Framework Papers

account_balance

From Dogs to Drivers of the Dow

Analyzing the evolution from one-variable contrarian value rules to multi-information systematic market leadership.

trending_up

From Averages to Algorithms

The history and mathematics of equity indexing from Charles Dow through Bogle and Arnott to modern direct indexing.

Intellectual Heritage

Pioneers of Indexing & Strategy

From statistical measurement and low-cost funds to smart beta and agentic cognitive systems.

psychology

Jon DuPrau

Agentic Systems · PrismVest

DuPrau leads the PrismVest cognitive architecture, treating the portfolio as an orchestrated network of specialized AI agents governed by human fiduciary judgment.

account_balance

John C. Bogle

Democratization · Vanguard

Bogle transformed the index into an investable strategy by founding Vanguard in 1974 and launching the first retail index mutual fund in 1976.

hub

Rob Arnott

Smart Beta · Research Affiliates

Arnott pioneered Fundamental Indexation, demonstrating that stock portfolio weights could be separated from price using sales, cash flow, and book value.