Executive Summary

Artificial intelligence (AI) has reached an inflection point.  Compute capacity has increased by more than 600 times in six years, while AI performance now exceeds human baselines on standardized assessments including the bar exam, LSAT, and GRE. ^[1] Roughly 95 percent of business leaders plan to increase AI investment in 2025, yet only 6 percent of U.S. companies have moved AI projects into production.^[2],[3]  This gap between conviction and execution defines the current AI landscape. This note presents a framework for understanding why AI is gaining traction and what distinguishes organizations that create value with AI from those that stall in perpetual AI pilots.  Drawing on production deployments across financial services, logistics, insurance, and manufacturing, the conclusion is that AI success is driven by execution discipline, not model sophistication.

Artificial Intelligence: The Shifting Landscape

Multiple forces are compounding to reshape the economics of enterprise technology.  A single NVIDIA B300 GPU now performs 14 quadrillion operations per second, more than 600 times the state of the art 10 years ago. ^[4]  The combination of falling costs and rising capability has expanded both the number of business problems solvable with technology and the depth at which existing problems can be addressed.  AI is not a linear improvement; it is an exponential shift that is eroding long-established competitive advantages.

Traditional approaches to automation and analytics are insufficient.  Rule-based systems cannot process unstructured language at scale.  Static dashboards cannot adapt to real-time operational conditions.  Manual analysis cannot match the speed or consistency that competitive markets now demand.  For senior business and technology leaders, the implications are significant:

• A 600 fold increase in compute capacity over six years has made previously infeasible workloads economically viable. ^[5]
• Fragmented data environments and legacy architectures create friction that slows AI deployment even when investment is available.
• Regulatory scrutiny and reputational risk concerns introduce governance requirements that many organizations lack the maturity and ability to address.
• The opportunity cost of delayed adoption compounds as competitors operationalize AI into core workflows.

Leading organizations recognize that AI is not a future consideration but a present competitive requirement.  The question is no longer whether to invest but how to move from investment to impact.

Why AI Adoption Still Lags: The Structural Barriers

Despite overwhelming executive conviction, the path to delivering value with AI is still not clear for most organizations.  Survey data reveals six structural barriers that prevent AI adoption^[6]:

1. Limited access to skilled resources (50 percent of those surveyed).  Execution-ready AI practitioners remain scarce.  Organizations that lack technical talent struggle to translate pilot projects into production systems.  The gap between AI experimentation and operational deployment requires cross-functional expertise that most enterprises have not yet built.  For example, a global logistics operator addressed this barrier by partnering with Andersen Consulting to deliver an integrated People Management Suite in eight weeks.  Supervisors now act in under two minutes on decisions that previously took 15–30 minutes.
2. Constrained by AI strategy and vision (43 percent of those surveyed).  Many organizations lack a coherent AI vision tied to business priorities.  Without clear strategic direction, AI initiatives proliferate without coordination, consuming scarce resources while failing to deliver measurable impact.  Executives recognize AI’s potential but struggle to translate that conviction into a sequenced roadmap.  For example, a leading U.S. life insurer overcame this barrier by narrowly focusing on reducing the number of life insurance policies being cashed in.  This insurer had $153 million in policies cashed in in Q1 2025 alone.  By narrowly focusing on this problem, they achieved double-digit percentage reduction in policies being cashed in within one quarter. 
3. Worried about regulatory compliance (38 percent of those surveyed).  Fear of regulatory exposure and reputational risk slows AI decision-making.  Organizations in heavily regulated industries face uncertainty about how AI-generated outputs will be treated by auditors and regulators.  This uncertainty creates paralysis, with legal and compliance teams blocking initiatives that lack clear governance frameworks.  For example, a global reinsurer addressed this barrier by implementing rule-based intelligence that automatically documented rationale and approvals for compliance, delivering production-ready value in approximately four weeks.
4. Not clear how AI drives value in the business (34 percent of those surveyed).  Leaders struggle to connect AI capabilities to concrete business outcomes.  The technology conversation often dominates, leaving executives uncertain about which problems AI actually solves and how to measure success.  This ambiguity makes it difficult to secure sustained investment or executive sponsorship.  For example, the Andersen Institute addressed this issue by building an AI platform that analyzed 1,500 earnings call transcripts and validated AI-extracted insights against realized margin outcomes demonstrating a direct linkage between AI capability and measurable economic intelligence.
5. Concerned about consequences of errors (32 percent of those surveyed).  Risk aversion prevents organizations from deploying AI into high-stakes operational environments.  The potential for AI to make costly mistakes, particularly in safety-critical or customer-facing contexts, creates hesitation even when the technology is ready.  Organizations default to manual processes they perceive as safer, even when those processes are slower and less consistent.  For example, a global industrial operator addressed this barrier by implementing predictive maintenance models with explainable, auditable recommendations, achieving a double-digit reduction in unplanned downtime.
6. Unsure how to measure costs and returns (29 percent of those surveyed).  Organizations lack mechanisms to quantify AI return on investment (ROI) and intangible value creation.  Traditional financial metrics do not capture improvements in decision quality, data trust, or operational resilience.  Without clear measurement frameworks, AI initiatives compete poorly for capital against projects with more established return profiles.  For example, a global reinsurance provider addressed this barrier by migrating investment analytics into Palantir Foundry with clear visibility into how data moved through the system, restoring business confidence in analytics and creating a durable platform for future AI use cases.

A Framework for AI Value Realization

Guiding Principles

AI is not magic.  It is another tool in the toolbox, but this tool is different for two reasons.  First, AI is increasing the number of business problems that can be solved with technology, expanding into domains like unstructured language and judgment-heavy workflows that were previously inaccessible.  Second, AI is deepening how existing problems can be solved, enabling more granular, real-time, and adaptive decision-making.  Both expose new pockets of value.^[7]

Value is realized when AI is treated as part of an integrated system and operating model, not a standalone initiative.  This principle shapes every element of successful AI programs: strategy is anchored to business outcomes, technology is selected to serve those outcomes, and execution embeds governance and change management from the start.  The core value proposition is then straightforward: AI enables organizations to solve problems they could not previously address and to solve familiar problems with greater speed, precision, and scale.

Foundation: Data Platforms and Governance

Sustainable AI programs are built on four foundational elements: unified data platforms that consolidate different and fragmented sources; governance frameworks that ensure auditability and compliance; modern infrastructure that supports real-time processing; and skilled teams that combine technical depth with domain expertise.  Without these elements, even sophisticated models fail to generate production value.

An example for a global reinsurance provider illustrates this pattern.  The client relied on an cloud-based investment analytics environment with fragmented transformation logic and inconsistent business rules.  Core datasets for securities, portfolios, positions, and holdings required extensive manual reconciliation before they could be trusted for downstream analysis.

Palantir Foundry served as the system of record for unified investment analytics.  Data sourced from external investment managers was standardized into a consistent, enterprise-aligned data model with clear traceability from raw inputs through transformed outputs.  The initiative was delivered over approximately 12–16 weeks.  Analysts accessed harmonized data without manual preprocessing, shortening analysis cycles by days and restoring trust in analytics used for capital and risk decisions.

Core Methodology: Three Pillars of Value Creation

Organizations that succeed with AI consistently organize their projects around three interdependent pillars: ^[8]

Pillar 1: Value Created.  AI initiatives begin with clearly defined business problems and economic outcomes.  Value is quantified before and after implementation.  Focus extends beyond cost reduction to enterprise value and intangible asset value, including data quality, decision velocity, and operational resilience.  Intangible assets have migrated from 17 percent of company value in 1975 to 90 percent today. ^[9] AI compounds the value of intangible assets like data, software, and industry expertise by removing low-value-add activities.  For example, at a large reinsurance company, a process that used to take two underwriters two weeks each, now takes as little as one hour through automated data gathering using AI.  It does this while simultaneously improving enterprise value through increased revenue, decreased cost, decreased risk, decreased time, and reduced capital requirements.

Work at the Andersen Institute has demonstrated this pillar in practice.  When the U.S. enacted a historic tariff increase in early 2025, the Institute needed to understand how companies were responding across costs, supply chains, pricing, and investment decisions.  The AI-enabled platform analyzed nearly 1,500 earnings call transcripts using a two-stage LLM classification pipeline that extracted 39 structured variables per transcript.  The system validated AI-extracted insights against realized margin outcomes, confirming that sectors reporting tariff-driven pressures were indeed experiencing greater margin declines.  Analyses that once required months now run in hours.

Pillar 2: Technology Applied.  Delivery relies on execution-focused teams combining deep technical skills and industry expertise.  AI is built on modern, governed data platforms that support scale, auditability, and speed.  Models and insights are embedded directly into operational workflows, not isolated dashboards.

Implementation at a leading U.S. life insurer demonstrates this pillar in practice.  In Q1 2025, $153 million in insurance policies were cashed in, driven by over 4,500 policy surrenders.  Associates struggled to navigate complex contract terms in real time, resulting in missed retention opportunities.  Andersen Consulting delivered an AI-powered Value Assistant built on Palantir Foundry and Artificial Intelligence Platform (AIP) that dynamically synthesizes customer and product data to generate context-specific talking points during live conversations.  For example, if a policyholder wants to cash out a $500,000 annuity with $150,000 in taxable gains, the tool shows tax implications and compares alternative liquidity options that avoid surrender charges.  A controlled pilot with 30 contact center staff indicated double-digit reduction in surrender-intent conversions.  The solution moved from concept to production in approximately 12 weeks.

Pillar 3: Execution Embedded.  Pilots are rapidly pushed into production.  Change management, governance, and adoption are embedded into delivery from day one.  Governance enables speed by creating confidence, not friction.

The experience of a global logistics operator demonstrates this pillar in practice.  The client managed thousands of frontline employees across multiple sites, with daily staffing decisions dependent on handwritten attendance logs and siloed HR systems.  Unplanned absences exceeded 8 percent on critical shifts, and supervisors spent hours each day reconciling attendance and coordinating replacements.

Using Palantir Foundry, Andersen Consulting unified the Human Resource Information System, time and attendance, and training records into a governed operational ontology.  Scoring models matched employees to open roles based on certifications, training history, and attendance patterns.  In Palantir AIP, supervisors managed staffing actions with clear rationale—reducing decision time from 15–30 minutes to under two minutes.  The solution moved from kickoff to production in eight weeks.  The client achieved double-digit reduction in unfilled critical roles, and time spent on manual reconciliation dropped by more than 30 percent.

Operationalization: From Pilot to Production

The gap between pilot and production is where most AI initiatives stall.  Organizations that close this gap share four characteristics:

• They assign executive sponsors with authority to remove organizational friction and allocate resources.
• They embed change management into delivery, training frontline workers before systems go live.
• They establish measurement frameworks that track both quantitative outcomes and workforce adoption.
• They build feedback loops that surface issues quickly and enable rapid iteration.

A global reinsurer illustrates this pattern.  The client managed thousands of active contracts across finance and procurement, supported by SAP for core financials and Coupa for spend and supplier management.  Contract processing relied on fragmented workflows, manual reconciliations, and offline coordination, extending cycle times and increasing operational risk.  Approval bottlenecks and limited end-to-end visibility delayed downstream financial actions.

Andersen Consulting mobilized a cross-functional team spanning finance operations, data engineering, and platform specialists.  Using Palantir Foundry, the team established a governed data foundation integrating SAP financial data and Coupa contract and procurement data into a single operational layer.  The solution identified stalled contracts, flagged approval exceptions, and surfaced required actions to the appropriate stakeholders.  When a contract exceeded predefined financial thresholds or deviated from standard terms, the system automatically routed it for additional review while documenting rationale and approvals for compliance. The end-to-end solution progressed from design to deployment readiness in approximately four weeks.  Contract approvals accelerated materially, reducing delays caused by manual handoffs and unclear ownership.  Finance and procurement teams eliminated redundant tracking and reconciliation, freeing capacity for higher-value work.  Leaders gained up-to-date insight into contract status, bottlenecks, and financial exposure across the portfolio.  Users trusted the data and workflows, improving decision quality and reducing rework.  “This initiative gave our teams clarity and control without adding complexity,” said the company’s Head of Finance Operations.  “We now move faster while strengthening governance across the contract lifecycle.”

Continuous Evolution: Sustaining Value Over Time

AI systems must evolve as business conditions, regulatory requirements, and competitive dynamics change.  Successful programs build adaptability into their operating models, treating model retraining and capability expansion as ongoing processes rather than discrete projects.  Integration with existing planning, budgeting, and governance cycles ensures that AI remains aligned with strategic priorities.

The experience of a global industrial operator highlights this pattern.  The client operated complex facilities with hundreds of critical assets, yet maintenance planning relied on manual logs, disconnected systems, and static inspection schedules.  Equipment failures routinely caused unplanned downtime.

Palantir Foundry consolidated asset master data, maintenance logs, inspection records, parts inventories, and sensor data into a unified system.  Predictive maintenance models used historical work orders and live sensor data to forecast failure risk.  When vibration and temperature readings crossed defined thresholds, the system flagged elevated risk and automatically generated a prioritized work order with required parts.  The solution was delivered in eight weeks.

The pilot achieved double-digit reduction in unplanned downtime for critical equipment.  Mean maintenance response time improved as work orders were triggered earlier and prioritized accurately.  “We now see problems before they disrupt operations,” said the company’s Head of Facilities and Engineering.

Strategic Implications

The shift from AI experimentation to AI execution creates five strategic imperatives for senior leaders:

1. Competitive Advantage.  Organizations that operationalize AI into core workflows will outperform those that treat it as peripheral.  The life insurer’s Value Assistant demonstrates this imperative with double-digit reductions in surrender-intent conversions translating directly into retained policy value, with early pilots stemming millions in potential outflows.  Speed itself is a competitive differentiator: the solution moved from concept to production in approximately 12 weeks.
2. Operational Excellence.  AI enables granular, real-time, and adaptive decision-making that traditional systems cannot match.  The industrial operator’s Facility Maintenance System shifted maintenance from reactive repair to predictive execution.  The logistics operator’s People Management Suite reduced staffing decision time from 15–30 minutes to under two minutes while achieving double-digit reduction in unfilled critical roles.
3. Risk Mitigation.  Governed AI programs reduce operational and regulatory risk by creating auditability, consistency, and transparency.  The reinsurer’s contract processing solution improved financial control while reducing cycle times with production-ready value was delivered in roughly four weeks without sacrificing governance.  All AI-generated outputs operated within governed data environments that enforce strict guardrails.
4. Talent and Culture.  AI augments human judgment rather than replacing it when implemented correctly.  Contact center associates armed with real-time, compliant guidance made better decisions faster, improving both customer outcomes and workforce confidence.  Logistics supervisors shifted from reactive coordination to proactive execution.  Maintenance technicians spent less time searching for information, increasing wrench time and consistency of execution.  Organizations that frame AI as a tool for human empowerment rather than human replacement accelerate adoption and reduce resistance.
5. Investment Priority.  AI investment must be treated as strategic infrastructure, not discretionary spending.  The Andersen Institute’s economic intelligence platform transformed nearly 1,500 earnings call transcripts into structured economic data replacing months of manual analysis with a repeatable, auditable capability.  The reinsurer’s investment analytics modernization created a durable platform for future AI use cases.  These investments created durable analytical capabilities, not one-time deliverables.

The common thread across these imperatives is execution discipline.  AI traction is driven by how organizations implement, govern, and operationalize, not by which models they select.

The Path Forward: From Conviction to Capability

AI capability will continue to advance rapidly, expanding both the range and depth of addressable business problems.  Organizations that have established governed data foundations and execution-focused operating models will capture disproportionate value as new capabilities emerge.  Those still building foundations will face widening competitive gaps.

“We are completely convinced the consequences of AI will be extraordinary and possibly as transformational as some of the major technological inventions of the past several hundred years: think the printing press, the steam engine, electricity, computing, and the Internet.” — Jamie Dimon, Chairman and CEO, JPMorgan Chase

Senior leaders should take four actions now: define AI strategy in terms of business outcomes rather than technology capabilities, assess data platform maturity as a prerequisite for AI deployment, establish governance frameworks that enable speed rather than create friction, and invest in execution-focused teams that combine technical depth with domain expertise.  The consequences of delay are pronounced: inflated operating costs relative to peers, missed productivity gains, and market loss to competition.

---

©2025 Andersen Institute for Finance & Economics. All Rights Reserved. This material is confidential intellectual property of the Andersen Institute for Finance & Economics. The views expressed in this note are those of the authors and do not represent an official position of The Andersen Institute for Finance and Economics or affiliated organizations.  By viewing this Andersen Institute Insight, you agree that you will not directly or indirectly copy, modify, record, publish, or redistribute this material and the information therein, in whole or in part. No warranty or representation, express or implied, is made by the Andersen Institute or any of its affiliates, nor does Andersen accept any liability with respect to the information and data set forth herein. Distribution hereof does not constitute legal, tax, accounting, investment or other professional advice. The information provided herein is not intended to provide a sufficient basis on which to make an investment decision. Recipients should consult their own advisors, including tax advisors, before making any investment.