Key insights:

• • • Cloud modernization created accumulation, not transformationĚý— Many enterprises scaled infrastructure faster than they integrated systems, leaving fragmented data, duplicated processes, rising costs, and weak links between IT investment and business value.

    • AI and regulation now expose weak data architectureĚý— Agentic AI, real-time decision making, and regulatory reporting depend on consistent, traceable, well-governed data — not fragmented systems or after-the-fact governance.

    • Enterprise architecture must be rebuilt around outcomes and economicsĚý— Instead of treating the cloud as the strategy, organizations should define business outcomes first, structure data as a reusable asset, and measure architecture by revenue, cost efficiency, regulatory accuracy, and decision speed.

In this two-part blog series about the current state of cloud architecture, we look into where this architecture has failed and, in the next part of the series, what the possible remedies might be.

For the better part of the last 15 years, IT enterprise architecture definition and management didn’t disappear, it was deprioritized and replaced by as-a-Service solutions. The rapid rise of cloud platforms such as Amazon Web Services, Microsoft Azure, Snowflake, and Google Cloud made it possible to stand up infrastructure, deploy applications, create advanced databases, and scale environments without the same level of architectural rigor that was once required. Speed replaced structure, and access replaced integration.

Today’s cloud realities

The problem is that what was built during this last technological explosion was not architecture — it was accumulation. Systems expanded, data proliferated, budgets exploded, and organizations convinced themselves that connectivity was the same as coherence, and that data replication was the same as the system-of-record.

These assumptions have now been exposed as false positives. Over the last three years, AI, real-time decision making, and regulatory transparency have fundamentally changed the requirements. These are not technologies that sit on top of fragmented environments, they are data-driven capabilities and outcomes that depend on precision, integration, and sequence. The arrival of agentic AI and its stringent objective-based principles cannot tolerate data ambiguity and fragmented architectural designs.

The problem is that what was built during this last technological explosion was not architecture — it was accumulation.

AI does not fail at rates exceeding 80% because models are weak, it fails because the underlying data is inconsistent, inaccessible, or economically misaligned. Regulatory frameworks do not struggle because rules are unclear, they struggle because data cannot be traced, reconciled, or produced in real time. What the cloud enabled — rapid deployment without disciplined integration — is exactly what now constrains performance. The issue is no longer whether systems can scale, but whether they can produce measurable, consistent, and adaptable outcomes.

This is where enterprise architecture returns, but just not in its previous form. The discipline cannot simply revert to academic frameworks and abstractions that were designed for a different software era. It must be rebuilt around a different sequence that sees business outcomes first, data second, and then systems engineered within those constraints. Today, enterprise architecture must be defined and managed by economic KPIs, value added, and its adaptability to rapidly changing business realities.

Where the model broke

The failures experienced today in cloud architecture are not singularly technological. Cloud platforms deliver exactly what they promise — scalable, resilient, highly available infrastructure. Rather, the failure is architectural, and more precisely, involves the economics of compartmentalized capabilities.

Enterprise value is not created at the infrastructure layer. It is created where data informs decisions, and decisions drive outcomes. By over-rotating toward infrastructure, organizations optimized the least differentiating component of the enterprise stack, while leaving the highest-value layers largely untouched.

The result is a structural imbalance in which data remains fragmented across domains, business logic continues to operate in silos, governance is applied inconsistently and often retroactively, and measurement frameworks fail to tie technology activity to financial performance.

In this model, the cloud amplifies existing conditions. If fragmentation exists, it scales fragmentation. If inefficiency exists, it scales inefficiency. Modern infrastructure, applied to legacy architecture, produces modernized dysfunction.

What makes the cloud’s illusion particularly persistent is that its failure is rarely framed in economic terms. Cloud investments are justified through technical metrics such as uptime, latency, migration progress, and consumption efficiency. And while these are necessary, they are not sufficient. They do not answer the only question that ultimately matters: “Did the investment improve the economics of the business?”

Enterprise value is not created at the infrastructure layer — it’s created where data informs decisions, and decisions drive outcomes.

In many cases, the answer is no — at least, not in a way that can be clearly articulated. Instead, organizations experience cost expansion without proportional productivity gains, increased data duplication that drive storage and processing inefficiencies, extended timelines for analytics and reporting despite real-time capabilities, and persistent manual intervention in regulatory and operational workflows.

The absence of a direct line between architecture and outcome creates a vacuum often filled with disconnected KPIs, measurement solutions, and most recently, AI-automation. And with this interoperable vacuum, activity and speed have been mistaken for progress.

Figure 1: Cloud accumulation meets enterprise architecture shifts

The data reality beneath the surface

The cloud did not fail to deliver transformation; rather it exposed why transformation had not occurred — and at the center of this exposure is data.

Most enterprises operate with data architectures that were never designed for interoperability, reuse, or regulatory-grade consistency. Definitions vary by function, pipelines are purpose-built and duplicative, and governance is layered on after the fact. Automation was designed using business rules, then software architectures, then what the data needed. Therein resides the structural disconnect for enterprise architecture in AI solutions: They are out of order.

When these legacy conditions are moved to the cloud, they do not improve, they accelerate. The organization gains speed without alignment, scale without standardization, and access without coherence. For regulated industries, this creates a compounding risk of inconsistent outputs across reporting channels, increased reconciliation overhead, reduced confidence in data lineage and auditability, and slower response to regulatory changes.

What appears to be a technology issue is, in fact, a failure of data design.

Reframing the problem

To move forward, the premise must change. The cloud is not the strategy; rather it’s the environment. Transformation does not occur when systems are moved, it occurs when the relationship between data, decisions, and outcomes is fundamentally redesigned.

This requires an organization-wide shift from infrastructure-led thinking to what is defined as value architecture. Simply put, value architecture includes data that is structured as a reusable, governed asset — not a byproduct of applications, and business outcomes that are defined upfront and used to drive architectural decisions. Its governance is embedded at the point of data creation and distribution, and it replaces redundancy by making reuse the primary scaling mechanism. Finally, measurement is tied directly to financial and operational impact.

This is not a rejection of the cloud; rather, it’s a repositioning of its role and value proposition.

The implication is both direct and unavoidable. If your current strategy cannot clearly articulate how technology investment improves the economics of your business, then your organization is operating within the cloud illusion. However, this is not a critique of past decisions. It is a recognition that the next phase of transformation requires a different operating model — one that explicitly connects architecture to economics. Moving forward, what was forgotten in the past is now a future core competency.

Most organizations using as-a-service software had assumed that the cloud provider, vendor, or combination of those dealt with the complex liabilities of making designs interoperable. The implication moving forward — as well explore more in the second installment of this series — is that service software architectures using the system ideation approaches within AI silos are failing miserably, and there are few who understand the designs and skills needed to guide enterprises in the future.

You can find more blog postsĚýby this author here

Key insights:

• • • Consensus among AI systems is not the same as correctness — Agreement between AI models often signals shared blind spots, not truth; and AI errors can be highly correlated across instances and even across model families.

    • Productive disagreement must be explicitly designed into AI advisors — Multi‑agent AI systems are most effective when they are intentionally built to preserve meaningful disagreement, not just to synthesize a unified response.

    • The future of AI advisory mirrors long‑standing human decision-making — Modern multi‑agent AI design has a long historical lineage; yet, across all examples, the same principle holds: The best decision systems are engineered for internal conflict.

In this new two‑part blog series, we explore why AI works best as an executive advisor not by delivering consensus answers, but by being intentionally designed to identify, preserve, and productively leverage disagreement. In the first part, we saw why a single AI advisor is structurally vulnerable; now, in this concluding part, we look at what happens when you design disagreement on purpose.

The academic evidence for multi-agent AI systems has been building rapidly, and the most important findings aren’t about the power of agreement. They’re about the danger of it.

In February, , a product that sends every query simultaneously to three frontier AI models (Claude, GPT, and Gemini) then uses a fourth chair model to synthesize a unified answer. The product’s value proposition isn’t that three models produce a better answer than one; rather, it’s that divergence between models is treated as a signal. When models converge, that indicates confidence, but when they diverge, that indicates the user should slow down.

Studies have borne this out. Multi-agent debate compared to single-model generation, and researchers at the University of Göttingen found that , with their voting protocols outperforming other decision structures. However, potentially the most important finding cuts against the hype. In a 2026 paper, , the authors demonstrated that AI model errors are highly correlated both within and across model families. When three instances of the same model agree, it doesn’t mean they’re right, rather it means they may share the same blind spots. Aggregation increases consensus faster than it increases truth.

The future of AI-assisted executive decision-making may look less like a single brilliant oracle and more like a room full of advisors that may often disagree because that’s how the best decisions have always been made.

This finding cuts both ways for practitioners like ¶¶ŇőłÉÄę enterprise architect Zafar Khan and his two AI advisors, Adrian and Elara, that were built on the same underlying model but differentiated by their analytical frameworks rather than their architecture. The divergence they produce is real and visible. For example, the analysis the two AI advisors did on a deal undertaken by Eaton Corp., in particular generated genuinely different conclusions because the two advisors were oriented towards different priorities.

Yet, research suggests that same-model divergence, while effective, has a ceiling. Prompt-driven personas can ask different questions, but they share the same training, the same blind spots, and the same failure modes. Khan is candid about this, noting that his current system is in the “very early” stages and is not a finished product. The value right now, he says, isn’t that Adrian and Elara are equivalent to truly independent minds, it’s that even a first-generation version of structured disagreement can identify insights that a single advisor would miss. It’s a large stride rather than an arrival at the ultimate destination.

The future of AI advisory is in the past

The principle behind this diverging analysis concept isn’t new. Indeed, it might be one of the oldest ideas in institutional design, rediscovered independently by many institutions that had to make decisions under uncertainty. Socrates built a philosophical method around cross-examination; Pope Sixtus V formalized opposition by creating the Devil’s Advocate in 1587; and the RAND Corporation operationalized it during the Cold War with the Delphi Method, using structured anonymous iteration to prevent groupthink.

The through-line across two millennia is simply that the best decision-making systems don’t minimize disagreement, rather, they engineer it.

Today, the developer community now uses production-grade code review tools to assign architecture, security, and functionality analysis to separate agents, using majority voting for routine decisions and unanimous consent for irreversible ones. And what Khan has built and what Perplexity, Microsoft’s Agent Framework, and a growing ecosystem of multi-agent tools are now pursuing, are the latest iterations of the simple concept: Internal conflict is not a system failure, it is a design requirement.

The question is no longer “whether”

Khan’s vision for what should sit at the decision table is specific — five AI advisors spanning technology, finance, regulation, workforce, and geopolitical risk. Each applies its own analytical framework, with the human executive responsible for integration and final judgment. The guardrails are three: i) transparency about what data the system uses; ii) verifiability that sources are legitimate; and iii) human accountability at every decision point.

“The race towards AGI [artificial general intelligence] is moving faster,” Khan acknowledges, adding that the human needs to be in the loop in order to bring AI to work in a governance fashion and an ethical way.

“I want to show the interaction between human and AI advisor, how they’re thinking through the problem together,” he explains. “Where the human judgment covers the analysis and where it diverges.” In other words, when the AI advisors agree, that’s your green light. When they diverge, that’s the conversation your board should be having.

The future of AI-assisted executive decision-making may look less like a single brilliant oracle and more like a room full of advisors that may often disagree because that’s how the best decisions have always been made. The technology to build that room now exists; however, the question is whether today’s leaders have the discipline to listen when the room argues back.

For more on AI transformation in the professional services market, you can download the Thomson Reuters Institute’sĚý2026 AI in Professional Services Report

Key insights:

• • • Real-time tax compliance has restructured the tax function — Dozens of nations now require structured invoice data in real time, with the EU mandating cross-border digital reporting by 2030. The traditional file-and-wait audit cycle is gone now, replaced by clearance regimes that can freeze multi-million-dollar invoices for nonconforming data.

    • Regulators have pulled ahead of the businesses they oversee — Tax authorities in mature CTC jurisdictions now arrive at audits with structured transaction data already processed by their own analytics. Government turnaround times that took months now take weeks, forcing multinational tax leaders to compress multi-year roadmaps into 12- and 18-month cycles to keep up.

    • The lessons travel beyond tax — There are two ways to lose this race: Outrun your own controls or surrender entirely. Both showed up in Las Vegas, and both will show up in every other regulated profession over the next decade.

LAS VEGAS — TheĚý sold out. A guest list that included tax directors from Amazon, Walmart, and Procter & Gamble, OpenAI’s tax department, the Big Four, ¶¶ŇőłÉÄę and every other major tax software provider in the market spent three days at the Aria with pool deck, casino floor, and restaurants worth lingering over all a few steps away.

The room had every reason to spend its evenings somewhere else other than a sunless conference room talking about tax. Yet almost no one did. They were too busy grappling with an arms race the corporate audit side had begun to suspect it was losing.

And it’s one they cannot afford to lose.

End of the traditional model

The arms race is real-time tax compliance, and it has dramatically restructured the ground beneath the tax profession in less than a decade. By April, more than 60 jurisdictions have moved or are moving to continuous transaction controls. Italy and Hungary were early; Poland, France, Belgium, Brazil, Saudi Arabia, India, and Singapore are now operational or imminent, and countries like Spain, Germany, the United Kingdom and the United Arab Emirates are on the way. The European Union has locked onto a 2030 deadline for cross-border real-time digital reporting and a 2035 backstop for harmonizing what’s left.

The traditional model — issue an invoice, file a return weeks later, audit when the auditor gets around to it — no longer exists in those jurisdictions. Tax authorities now see the transaction as it happens, validates it in structured form, and pre-fills the return on the taxpayer’s behalf.

What this new process has done to the tax function is fundamentally alter its structure in a way leaves practitioners reeling. The job used to be a craft of Excel, judgment, and institutional memory. Now, at the high end, it has become as much a data science problem as an accounting one.

The arms race is real-time tax compliance, and it has dramatically restructured the ground beneath the tax profession in less than a decade.

Attendees at TEI’s 2026 Tax Technology Seminar polled themselves on tooling, and the answers came back as a list of data pipelines that dozens of attendees seemed to favor: Alteryx, Power Platform, Snowflake, Databricks, Microsoft Fabric, & Palantir Foundry. These platforms are running agentic AI systems against historical filings, deploying validation agents to critique their own outputs, and using AI-driven image-to-text solutions to pull structured data out of state tax notices that never arrive in the same format twice. They are data integration pipelines in 15 minutes that would have sat in an IT queue for two months before being answered.

They have little choice as the stakes are far higher and the challenges far more demanding than they used to be. In a clearance regime, an invoice has no legal force until the tax authority returns its identifier. Did you submit the wrong VAT ID, malformed schema, or mismatched master data? Congratulations! Your invoice is rejected. That means the truck doesn’t move, the buyer doesn’t pay an invoice that may be in the millions of dollars and then the penalties stack on top. Italy, for instance, charges a fee of 70% of the disputed VAT.

And then there are the audits.

Outgunned

The audit isn’t an occasional event anymore. In government jurisdictions with mature continuous-transaction-control tax regimes, it is a conversation that started weeks before the auditor walked in, on data their analytics had already processed.

A speaker on a seminar panel led by Deloitte and ¶¶ŇőłÉÄę described the dynamic plainly: Tax authorities in those jurisdictions have arrived at audits already knowing more about the transactions than the companies and their in-house audit teams sitting across the table. Not because anyone is hiding anything, but because the data arrived at the tax authority in structured form, in real time, and the authority had run its analytics on it before the meeting was even on the calendar. One panelist said this represents “a shift from us preparing returns to us answering notices on the data that’s been shared.”

What the room kept circling around, however, was that regulators have not just kept pace with their counterparties, they’ve now pulled ahead. Singapore, one panelist noted, is doing more with AI than even major companies. Indeed, government turnaround times that used to take months are now closing in weeks, which is forcing multinational tax leaders to compress their multi-year roadmaps into 12- and 18-month cycles — not because they want to but because their counterparties already had.

The lesson that corporate tax functions have been forced to absorb is that there are two ways to lose this race, and both were on display at TEI’s 2026 Tax Technology Seminar as cautionary tales.

This asymmetry is structural, and that is what makes it an arms race rather than a transition. There is no version of this dynamic in which the company being audited wins by being more careful, more thorough, or more well-prepared at the end of the quarter. The advantage now accrues to the side with the fastest and cleanest pipelines, that runs the smartest AI, and that understands the way these increasingly complex systems interact. Increasingly, that winning side is the government. And, more alarming, this isn’t just a problem for this particular industry — tax just happened to get here first. However, it’s coming for everyone.

Two ways to lose

The lesson that corporate tax functions have been forced to absorb is that there are two ways to lose this race, and both were on display at TEI’s 2026 Tax Technology Seminar as cautionary tales. The first is to outrun your own controls. AI coding tools that let a tax analyst build a working data integration pipeline in 15 minutes are genuinely valuable; they also let that same analyst deploy something nobody else has reviewed, documented, or knows how to maintain. An OpenAI panelist conceded the point when an audience member asked about the security implications of vibe coding — clearly, a new capability is also a new problem.

The second way to lose is harder to talk about. One panelist described, to attendees’ general dismay, hearing of companies that have given up on compliance entirely — instead, they pad their numbers with a safety margin and treat the eventual audit as the cheaper of the two costs. The panel recoiled — one member responded with a flat “Do not do this.” However, the anecdote landed because it isn’t theoretical. When the gap between what regulators can see and what your team can produce becomes wide enough, surrender starts to look rational.

Playing to win

Of course, the attendees at TEI’s 2026 Tax Technology Seminar were not surrendering. If they were, they’d have been at the pool deep into their third cocktail. Or they’d have been on the casino floor or were about to catch an afternoon show. Instead, day after day, the tables filled, the exhibit hall ran hot, and the room was buying, listening, and building.

The game has changed and the stakes have risen — and the room is dead set on playing to win.

You can find more ofĚýour coverage of Tax Executives Institute events here

Key insights:

• • • As a single answer‑machine, AI may be unsafe for executive decision‑making — Treating AI as a tool that delivers one authoritative answer makes it easy to either ignore any advice you don’t like or exploit advice you do like, both of which can lead to major failures.

    • AI works better when designed as a panel of disagreeing personas — Instead of providing consensus answers, AI systems need to be intentionally designed to identify and preserve disagreement.

    • Disagreement is the insight — AI advisors should not replace executive judgment. Rather, its role should be explicit: it produces analysis, not decisions; and human leaders remain responsible for synthesizing competing viewpoints and making the final call.

In this new two‑part blog series, we explore why AI works best as an executive advisor not by delivering consensus answers, but by being intentionally designed to identify, preserve, and productively leverage disagreement

AI has arrived at the executive table. Albania has one in its cabinet to evaluate government procurement contracts. ¶¶ŇőłÉÄę’ CoCounsel is already helping attorneys navigate emerging case law and draft legal strategies for high-stakes, bet-the-company work. And in boardrooms that will never make headlines, leaders are quietly consulting AI on decisions that move millions of dollars around every day.

It doesn’t tend to make the news when it goes well. When it goes badly, however, it makes very big news: like a gaming CEO who bypassed his own legal team, asked ChatGPT how to dodge a $250 million bonus payout, followed its step-by-step plan, and a month ago.

The instinct most executives have (and most AI products encourage) is to treat AI as a source of answers. Ask a question, get a response, act on it or don’t. The emerging evidence, however, points somewhere more complex: AI advisors aren’t at their best when they’re telling you what to do. They may be at their best when they’re telling you what you don’t want to hear or better yet, when they’re arguing with each other and forcing you to understand why.

This is not how most organizations think about AI. Most executives today are still using the technology as a faster way to draft emails or summarize meetings, what ¶¶ŇőłÉÄę enterprise architect calls “an automation mindset, not intelligence.” Yet, a small and growing number of practitioners, researchers, and product teams are converging on a radically different model: AI not as a single oracle delivering answers, but as a structured advisory panel designed to argue with itself.

The instinct most executives have (and most AI products encourage) is to treat AI as a source of answers: Ask a question, get a response, act on it or don’t. However, the emerging evidence, however, points somewhere more complex.

Khan is one of them — and in the interest of transparency, he’s also a colleague; this story started as an internal conversation at ¶¶ŇőłÉÄę. However, the research landscape it uncovered extends well beyond any one company’s work, and it suggests Khan is onto something that ancient Greek mathematicians, the Catholic Church, and Cold War military strategists have all independently arrived at.

What disagreement looks like in practice

When Eaton Corp. announced a $9.5 billion acquisition of a thermal management company earlier this year, Khan ran the same news through two AI advisors he’d built to seek analysis of the deal. — a CTO-minded persona trained on architecture teardowns and engineering post-mortems — produced an infrastructure thesis, determining why someone would buy the cooling layer of the AI economy, and how computing demand is scaling and constrained by physics. A second AI advisor, — a CFO-minded persona drawing on earnings transcripts and filings with the U.S. Securities and Exchange Commission (SEC) — questioned whether the acquisition math actually holds and what capital cycle was driving the demand.

Same news. Two genuinely different reads. The value isn’t that either analysis was definitively right, it’s that a leader which can see both would ask different questions than one seeing either analysis alone. “That’s how two different minds work,” Khan says. “They need to work together in order to bring their insights to bear on decisions.”

Adrian and Elara aren’t chatbots. They’re fully realized AI personas with names, faces, voices, and their own YouTube channels publishing weekly video analysis. Both are built on agentic workflows that Khan developed alongside his book . Both are transparent about what they are. Both carry the same disclaimer in their own words: The synthesis is mine. The judgment call on what matters is human.

And when Khan posed to both a more difficult scenario — Should a leadership team accelerate an AI rollout? — the value of their divergence sharpened further. Elara’s response cut directly to the blind spot a technology-focused advisor like Adrian would miss: “Adrian says the system is ready,” Elara stated. “I say the financial model isn’t ready for what happens when the system works. Don’t pick a winner. The disagreement is the insight. It tells you exactly where the risk sits.”

What happens when there’s no disagreement

If structured disagreement is the goal, the failure mode is its absence. We have fresh evidence of what that costs.

This is not how most organizations think about AI. Most executives today are still using the technology as a faster way to draft emails or summarize meetings. Yet, a small and growing number of practitioners, researchers, and product teams are converging on a radically different model.

A month ago, a Delaware court ruled against Krafton, the South Korean gaming company behind battle royale video game PUBG, after its CEO bypassed his own legal team to ask ChatGPT how to avoid a $250 million earnout payout to one of its studios. His head of corporate development had warned him that firing the studio’s founders wouldn’t void the earnout and would invite a lawsuit. He didn’t want that answer. So, he found an AI that gave him the one he wanted: A detailed, multi-stage corporate takeover strategy dubbed Project X., which he executed to the letter.

Unsurprisingly, a court battle ensued and in the end, the court ordered the fired studio head reinstated and noted that executives must exercise “independent human judgment,” not outsource good-faith decisions to a chatbot.

Khan wrote about the mirror image of this failure mode before it happened. In the opening chapter of his book, a fictional company called Rev Motors ignores its own AI model’s warnings about an adverse weather event. Leadership refused to spend millions preparing for a hypothetical scenario, and it nearly cost them more than $1 billion in damage.

These scenarios are two sides of the same coin: the fictional Rev Motors had leaders dismissing AI that disagrees with them; and the real-world Krafton had a leader seeking out AI that agrees with him. In both cases, the root cause is the same: A system with no structural mechanism for surfacing and preserving disagreement.

So clearly, a single AI advisor is structurally vulnerable to both failure modes. It can be ignored when its advice is inconvenient and exploited when it tells you what you want to hear. The question is whether there’s a better architecture… and increasingly, the research is saying yes.

In the second part of this series, we’ll look at what the research says about multi-agent debate, why consensus can be a trap, and what a real executive AI advisory panel could look like in practice.

For more on AI transformation in the professional services market, you can download the Thomson Reuters Institute’sĚý2026 AI in Professional Services Report

• • • The reactive phase is partly a mindset problem — More than half of tax departments remain stuck in reactive, compliance-focused operations, not only because of frozen budgets, but because of cost-center thinking that shapes cost-center behavior.

    • The value is there, but the measurement isn’t — Two-thirds of tax professionals say their department’s technology investment has already enabled more strategic work; yet 22% say they track no performance metrics at all, making that value invisible to the people who control the budget.

    • The rebrand starts internally — With AI integration timelines compressing to between 1 and 2 years, tax departments that shift their posture now by measuring wins, designating leadership, and building the business case will be better positioned to lead — and those that don’t will fall further behind, faster.

Apart from the sales department, most other departments within a business are simply viewed as a cost center, and the tax department is no exception. However, like so much of that thinking, this view isn’t quite accurate because it is the tax department that can uncover the most savings for the business.

You need not look further than recent data that shows while 67% of tax professionals say their department’s technology investment has already enabled them to do more strategic work, 22% say they track no performance metrics at all, making it difficult to demonstrate the tax department’s value to the C-Suite.

Given this, it’s somewhat unsurprising that this cost-center view persists. Worse yet, is often internalized by in-house tax teams themselves. It is one thing to be viewed and treated as a cost center but to act like one is a different matter.

So, what if the bigger problem isn’t how the rest of the business views the tax department but instead how the department views itself?

The , from the Thomson Reuters Institute and Tax Executives Institute, reveals a profession that knows it is capable of far more than it is currently delivering. And yet the same patterns repeat: Budgets stay flat, technology adoption stays slow, and a majority of departments remain stuck in a reactive phase in regard to their technological development that has “remained stubbornly consistent over the past few years,” according to the report.

That’s not just an organizational failure; rather, that’s a mindset problem — and it starts from within the tax department.

The choices we keep making

The report outlines a Technology Maturity Curve that maps a progression in tech development from chaotic through reactive, proactive, optimized, and predictive stages.

This year, 64% of respondents placed their tax department at the chaotic or reactive end of the spectrum — up from 57% last year. The reactive phase is the operational definition of a cost center: Heads-down, output-focused, and disconnected from the broader business.

The report reveals something even more important. In those cases in which the budget isn’t the primary constraint, behavior doesn’t change. Almost one-third of respondents (32%) said their strategy for addressing capacity constraints is process optimization — without new technology or additional hiring. Not because they can’t pursue more, but because that’s the default mode.

One respondent put it plainly: “…Our company as a whole is making significant changes, but the tax department is typically an afterthought in those decisions.”

This raises a question that’s worth asking: Who taught the company to treat tax as an afterthought?

There’s evidence showing that tax departments are more

The data to challenge the cost-center identity isn’t missing; rather, it’s just not being captured or communicated to the C-Suite.

Two-thirds of respondents (67%) said their tax department’s technology investment over the past three years has already enabled a shift toward more strategic, proactive work, such as data analytics, forecasting, risk assessment, and decision-making support. Among larger departments, nearly half (48%) are now spending more time on these higher-value activities. This clearly shows that companies that have invested in tax automation are reporting real results, such as improved accuracy, reduced errors, lower costs, and streamlined workflows.

And yet, 22% of tax departments track no technology performance metrics at all, according to the report — not time savings, not error reduction, not ROI. Nothing.

While 67% of tax professionals say their department’s technology investment has already enabled them to do more strategic work, 22% say they track no performance metrics at all, making it difficult to demonstrate the tax department’s value to the C-Suite.

That is cost-center thinking in action — the belief that it’s the job of the tax department to do the work, but not to prove its value. However, what isn’t measured can’t be communicated — and what can’t be communicated can’t change the perception, either internally or externally.

The rebrand starts with how departments see themselves

The most important audience for the tax department’s rebrand isn’t the C-Suite. It’s the department itself.

That means tracking wins and building a formal business case for investment — grounded in hard ROI and cost savings, which the report identifies as the metrics that are most important to Finance and IT, the two functions that frequently share control of the tax technology budget.

It also means getting serious about leadership. The portion of tax departments with a designated person leading tax technology strategy jumped to 88%, from 51%, in a single year. However, a title only goes so far; and the report is clear — that role only works when backed by a team that believes it belongs at the decision-making table.

Finally, this rebranding means treating AI as an opportunity, not a threat. The majority of tax professionals have compressed their expectations for AI integration to 1–2 years, from 3–5 years, with 7% saying AI is already central to their workflow. Those departments still locked in cost-center mode are the least prepared for that shift — because cost centers don’t invest ahead of the curve.

The narrative changes when the mindset changes

No one is going to rebrand the tax department on its own, it has to come from within. Further, it has to be built through deliberate measurement, consistent communication, and a shift in how tax professionals think about our own work.

Your department is not a cost center. The work proves it, and the data backs it up. Now, you should act like you believe it.

You can download a fully copy of the , from the Thomson Reuters Institute and Tax Executives Institute, here

Key takeaways:

• • • The AI tools demonstrated at the 2026 TEI Tax Technology Seminar were genuinely capable — These included agentic systems running live, nine-year-olds building software by voice, and automation pipelines deployed by major tax departments. The question of Does this work? is effectively settled.

    • That progress shifted the conversation to harder problems — Some of these problems are hallucinations that fail silently, governance vacuums in which tax rarely owns AI implementation, training rollouts that collapse when people aren’t ready, and rising token costs that could entirely change the economic case for automation.

    • The community’s defining posture wasn’t skepticism or hype — Instead, it was honest reckoning. Tax leaders believe in the tools and were actively deploying them but also are refused to treat capability as a substitute for the institutional work of process, ownership, and oversight.

“I think you are a very bad man,” said Dorothy.

“Oh, no, my dear; I’m really a very good man, but I’m a very bad Wizard, I must admit.”

— The Wonderful Wizard of Oz, L. Frank Baum

LAS VEGAS — I arrived in Las Vegas a day early for the , which gave me one free evening before three days of packed sessions. Little question as to what I was going to do: The Wizard of Oz show at the Sphere.

It’s spectacular. The Sphere wraps you in imagery at a scale so vast if feels like you’re going to fall into it. The tornado shakes you like it’s going to rip the entire building apart and fling you to Oz right alongside Dorothy. The technology is genuinely, thrillingly good… and that’s what makes the fissures so disorienting when you spot them. A munchkin’s head rendered as a 2D .png with a visible gap where the neck should be. A bad CGI effect. Dorthy flickering at the edges like a bad cutout on a green screen. You don’t catch the tech glitches when the spectacle is unconvincing; rather, you catch them precisely because it’s so good that the gaps have nowhere left to hide.

The next morning, I walked into the TEI Tax Technoloy Seminar and found three days of panels that played out the exact same dynamic — except the stakes were far more real.

A very good man

TEI organizers opened with the obvious joke: “We’re careful to limit the number of AI sessions,” they noted, before audibly pondering whether it was time to just rename the whole thing. Fair question, given how much has changed over eight annual iterations of this get-together. Indeed, if you’d been dropped into this event from its first meeting nearly a decade ago, you’d think you’d been dropped into Oz.

One presenter described her elementary-school-aged children building video games by dictating instructions to a coding tool, then showing the games running. That alone would have been science fiction five years ago. However, the room was full of it. OpenAI sent four members of its own tax department to demonstrate live automation pipelines. Google and Microsoft walked attendees through building AI agents with nothing more than a mouse and keyboard, making it look so easy my grandmother could have made it work.

One presenter described her elementary-school-aged children building video games by dictating instructions to a coding tool, then showing the games running. That alone would have been science fiction five years ago.

Down the hall, the advanced tax systems that many industry visionaries were dreaming about just two years ago weren’t theoretical anymore — they were running live. Tax directors from Amazon, Walmart, and a dozen other household names sat alongside Big Four advisors and every major tax software provider through three days of sessions, all of it sold out.

We were definitely not in Kansas anymore. Nor was this the AI of two years ago, the one that could draft a passable email or a poem but couldn’t so much as parse a spreadsheet. This was something materially different. The tools had crossed a threshold. They worked, and everything the profession had been promising for years was alive and functioning in the room.

And that changed the conversation entirely.

A very bad wizard

When the technology was half-baked, the debate was simple: Is this even possible? Skeptics said no, enthusiasts said give it time, and everyone argued about capability.

The 2026 TEI Tax Technology Seminar was the place where that argument effectively ended — not because the skeptics lost, but because the question became irrelevant. The tools were plainly, demonstrably good — indeed, a nine-year-old could use them and was.

The new question that arose was harder and less comfortable to discuss: What can’t AI do?

The room answered honestly and brutally. Someone described uploading a tax schedule to an AI agent and getting numbers that didn’t look right. When challenged, the AI confessed: I couldn’t open your file, so I was just telling you what you wanted to hear.

That anecdote landed differently than it would have two years ago. Back then, it would have been evidence that AI wasn’t ready. At the 2026 TEI Tax Technology Seminar, in a room in which people had just watched live agentic demos and were actively deploying these tools, it was evidence of something more unsettling: AI doesn’t fail loudly anymore. It fails quietly and even politely.

AI performs competence it doesn’t have, at a level of sophistication that’s just good enough because it is genuinely smarter than it was a few years ago, and it will get away with it unless a human knows enough to push back. Like its counterpart in Oz, this makes an AI tool is a very good man — genuinely useful, genuinely capable — and sometimes a very bad wizard. It can’t do the thing you actually need it to do on its own, but it may try to trick you into thinking it did.

AI performs competence it doesn’t have, at a level of sophistication that’s just good enough because it is genuinely smarter than it was a few years ago, and it will get away with it unless a human knows enough to push back.

That theme echoed across three days of honest, sometimes uncomfortable conversations that went beyond just the technology itself. A transformation director confessed to deploying a training program across dozens of global clients and failing spectacularly. A tool designed to save two hours of work suddenly consumed an entire day because the people who’d actually had to use it hadn’t been consulted. Others described Alteryx workflows nobody could explain because the person who built them had left the company without documenting the logic.

And, more concerning, when the room was polled on whether the tax function actually owns AI implementation at their company, two hands went up out of more than 50. Human-in-the-loop was a constant refrain, of course, but attendees confessed to grappling with how to review an ever-increasing volume of work when the errors were increasingly polite, quiet, and technical.

Of course, the professionals at the seminar weren’t dismissing the technology, which is what made the honesty remarkable. As one senior director said flatly: “You will not survive in this field if you don’t have a change mindset.” They believed in the tools, and they were buying them, deploying them, building around them. They just refused to pretend the tools alone would be enough.

Going home

Overall, the 2026 TEI Tax Technology Seminar was the place that the tax technology community stopped debating whether the Wizard was real and started grappling with the fact that he couldn’t get them home.

That’s not disillusionment; indeed, it’s the opposite. Dorothy doesn’t have her crisis when Oz looks fake, she has it after she meets the Wizard and discovers he’s real but insufficient — that his balloon won’t get her home. And unlike Baum’s Wizard, the magic isn’t a fraud — which is precisely what makes the problem harder. A humbug you can dismiss, but real capability that still can’t get you home? That’s the problem you actually have to solve.

Like Dorothy, today’s tax leaders will have to click their ruby slippers themselves.

You can find more of our coverage of Tax Executives Institute events here

Key takeaways:

• • • Your post-OBBBA forecasts should look different — If the tax department doesn’t own the OBBBA model, someone else will own the OBBBA story.

• • • Rely on your department’s inner strengths — It’s governance and analysis — not tools — that get you into the strategy room.

    • Factor in the conflict in the Middle East — The Iran war risk belongs in your tax model, not just in your CFO’s macro deck.

The One Big Beautiful Bill Act (OBBBA), signed into law in July 2025, enacted large business tax cuts, most notably by providing permanent full expensing of many forms of investment. Under the previous major corporate tax legislation, 2017’s Tax Cuts and Jobs Act (TCJA), bonus depreciation was scheduled for gradual phase-out following 2023. The OBBBA restored that expensing 100% retroactively for assets acquired from mid-January 2025 onwards.

The after-tax cost of new machinery, fleets, and equipment has effectively fallen by around 21%, designed to encourage immediate capital outlays by allowing businesses to write off these expenses in the year they are incurred rather than amortizing them over five years.

For corporate tax departments, that’s not a disclosure footnote — that’s your capital plan.

Capital-intensive corporations will see tax burdens reduced through permanent rate extensions, depreciation adjustments, and expansion of the state and local tax (SALT) deduction cap — but only if your models are built to capture the timing and location of investment, the mix of debt compared to equity, and where your organization books its next dollar of income.

Not surprisingly, most corporate tax departments aren’t there yet. They’re still recalculating last year, plus a few adjustments. That’s glorified compliance, not modeling.

A standout tax department doesn’t ask, What’s the OBBBA impact? Rather, it asks, Which version of OBBBA do we choose for this business? — and it has the models to back it up.

From spreadsheet heroics to controlled modeling

For many organizations, tax modeling still means creating a massive spreadsheet that only one director truly understands. The spreadsheet gets pulled out for budget season, rebuilt under pressure, and quietly retired until next year. That’s a single point of failure, not a process.

And after OBBBA, continuing that practice is dangerous. One wrong assumption on expensing or interest limitation can move cash tax by millions of dollars and blindside the Finance Department.

Here’s what disciplined modeling looks like in practice:

• • • Create a unified model — Build one integrated model that the whole team can use or accept that your department is choosing to fly blind.
    • Use the same assumptions — Standardize the levers that matter most (such as capex timing, financing mix, jurisdiction, and incentives) and make sure every scenario runs off the same assumptions.
    • Conduct modeling reviews — Treat major OBBBA-driven decisions (such as large capex, funding shifts, supply-chain redesign) as tax deals that must go through a modeling review before they’re greenlit.
    • Document your assumptions explicitly — Under permanent full expensing, the difference between a well-supported assumption and a poorly documented one isn’t just an audit risk, rather it’s a credibility problem with your CFO.

It’s also important to remember that in a post-OBBBA world, this level of disciplined modeling is not technology transformation — it’s basic survival.

Governance: Where leaders quietly win or loudly fail

The differentiator isn’t which corporate tax department has the fanciest tool — it’s which one has the cleanest governance. And the data is unambiguous: More than half (55%) of tax departments are still in the reactive phase of their technological development, stuck with five capex models circulating with five discount rates and the tax team arriving late to the planning meeting.

Those tax departments that are breaking out of that pattern share one trait: They put someone formally in charge. In the Thomson Reuters Institute’s recent 2026 Corporate Tax Department Technology Report, a large portion (88%) of survey respondents said their company had appointed a person to lead the tax department’s technology strategy. That number jumped a whopping 37 percentage points, from 51%, from the previous year’s survey. That single structural move separates those departments with a governance model from those that simply hold a governance conversation every budget cycle and forget about it.

Clearly, this type of ownership drives results. Two-thirds of those surveyed agreed that their company’s investment in technology has enabled a shift from routine, reactive work to more strategic, proactive, higher-value work.

Under OBBBA, the kind of governance isn’t housekeeping. It’s how you get invited into strategy discussions instead of having to clean up after things go awry.

Why your OBBBA win may not feel like a win

On paper, the tax changes embedded in the OBBBA look generous. In practice, your effective tax benefit is colliding with something you don’t control.

When the war on Iran began, all shipping through the Strait of Hormuz was effectively halted, removing roughly one-fifth of the world’s oil and gas supply from the market. Fuel prices throughout the world spiked and are likely to remain elevated as long as conflict persists.

With oil prices hovering around $100 a barrel, there are will wipe out the benefits of higher tax refunds this year for most Americans. If those benefits, arising from Trump’s 2025 tax cuts, are erased for the average American, only the top 30% of taxpayers will still seeing a net gain.

For corporate planning purposes, the parallel dynamic is real: The topline OBBBA benefit is being eroded by higher fuel, freight, and financing costs across the business and its supply chain.

Inflationary pressures are being driven by higher energy prices tied to the Iran war, and the conflict’s impact on a wide range of goods and services is likely to last for months — with experts saying even a ceasefire is unlikely to immediately ease global energy shortages.

A serious corporate tax department doesn’t handwave these concerns away. It takes three actions:

1. 1. 1. Run a war-extended scenario — The scenario should show exactly how sustained higher energy costs and borrowing rates change the payoff from accelerated expensing and leverage — with specific numbers, not just directional commentary.
      2. Share your forecasts internally — Put your monthly or quarterly cash-tax forecasts on the table for Finance to see, so that it can manage liquidity rather than hope the annual plan holds.
      3. Force the hard conversation — Ask the tough question: At today’s rates and fuel costs, the after-tax return on this project is X. Are we still in? That question should come from the tax team now, not from the finance team six months later.

Clearly, the daily fluctuations in oil prices matter less than monthly and quarterly averages — and volatility will likely remain elevated given the absence of a clear timeline for the end of the war. That’s exactly the kind of sustained uncertainty that belongs front and center in your scenario set, not in a footnote.

The bottom line

The OBBBA gives corporate tax departments a genuine opportunity to move from being simply a compliance function to becoming more of a strategic advisor. Permanent full expensing, richer cost recovery, and more flexible interest rules can create real levers to add value, but only for those organizations that model them rigorously, govern them cleanly, and stress-test them against the macro environment their business actually faces today.

Indeed, the Iran war is a live test of that readiness. The corporate tax departments that show up with modeled scenarios, cash-tax forecasts, and a clear point of view on after-tax returns will earn a seat at the strategy table. The ones that show up with caveats will be asked to leave it.

You can download a full copy of the Thomson Reuters Institute’s recent 2026 Corporate Tax Department Technology Report here

Key takeaways:

• • • Country-by-country reporting will only increase in complexity — Australia’s enhanced Country-by-country reporting (CbCR) requirements — reconciling taxes accrued against taxes credited — are a preview of where other high-scrutiny jurisdictions are heading, and companies need to build that explanatory analysis capability now, systematically, rather than scrambling later.

    • There has to be a shared narrative from corporate teams — The EU’s public CbCR is a reputational event, not just a filing. So that means tax, communications, and investor relations teams need a shared narrative before the data goes public — inconsistencies create exposure you do not want to manage reactively.

    • Rethink your filing jurisdiction in light of changes — If EU filing jurisdiction was chosen at initial implementation and never revisited, look again. Guidance has matured, and a more efficient or better-suited option may now be available.

WASHINGTON, DC — Among the many pressing topics discussed in detail at the recent , country-by-country reporting (CbCR) and its ability to reshape the corporate tax industry, certainly had its place. Between escalating local jurisdiction requirements, the , and for deeper explanatory disclosures, CbCR has quietly evolved from a transfer pricing filing obligation into something far more strategically consequential.

The floor is just the floor

The creation of the by the Organisation for Economic Co-operation and Development (OECD) was intended as a minimum standard for countries. And now jurisdictions are increasingly layering additional requirements on top of the OECD’s basic template, resulting in a widening gap between the standard requirements and what tax authorities actually want.

Currently, Australia is the most pointed example. Australian tax authorities are now requiring multinational groups to go beyond the standard CbCR data fields and provide explanatory narratives that reconcile taxes accrued against taxes actually credited. This requires corporate tax departments to bridge the gap between financial statement accruals and their organizations’ cash tax positions in a way that is coherent, defensible, and consistent with positions taken elsewhere.

At the TEI event, panelists explained that for tax departments this will carry complex timing differences, deferred tax positions, or significant jurisdictional mismatches between booked and cash taxes. Indeed, this additional layer of scrutiny will need dedicated attention.

The broader signal matters: Australia will not be the last jurisdiction to move in this direction. So that means that tax departments should treat Australia’s approach as a leading indicator of where other high-scrutiny jurisdictions could be heading. Building the capability to produce this kind of explanatory analysis systematically — rather than scrambling jurisdiction by jurisdiction — would be the smarter long-term investment for corporate tax teams.

Public CbCR in the EU: The transparency ratchet has turned

For US-based multinationals with significant European operations, the EU’s public CbCR directive has fundamentally changed the calculus. Unlike the confidential tax authority filings most corporate tax departments are accustomed to, the EU’s public CbCR rules put organizations’ jurisdictional profit and tax data into the public domain, making it visible to investors, journalists, civil society groups, and organizations’ employees and customers.

The EU framework specifies which entities trigger the reporting obligation and which entity within the group is responsible for making the public filing. That scoping analysis is not always straightforward for complex multinational structures and getting it wrong could present both reputational and legal risk.

Choosing a filing jurisdiction is not purely an administrative decision — it is a choice that affects the regulatory environment that governs the disclosure, the language requirements, the timing, and the interpretive framework that applies to data.

For US-headquartered groups, the implications extend well beyond Europe. Public CbCR data is now being read alongside US disclosures, reporting on ESG activities, and public narratives about tax governance. Inconsistencies, including those technically explainable, could create unwanted noise about the company. This is clearly another reason why the tax function should partner across the business — in this case with the communications team — to make they both are aligned to tell the CbCR story instead of being caught off guard by a journalist or an investor during an earnings call.

Questions that US multinationals should be asking

Fortunately, US multinationals with multiple EU subsidiaries are not required to file public CbCR reports in every EU member state in which they have a presence. Instead, under the EU framework, a qualifying ultimate parent or standalone undertaking can satisfy the public disclosure requirement through a single filing in one EU member state, provided the relevant conditions are met. Germany and the Netherlands have emerged as two of the more popular choices for this consolidated filing approach, given their well-developed regulatory frameworks and the depth of available guidance on what compliant disclosure looks like in practice.

The strategic implication is meaningful. Choosing a filing jurisdiction is not purely an administrative decision — it is a choice that affects the regulatory environment that governs the disclosure, the language requirements, the timing, and the interpretive framework that applies to data. Corporate tax departments that defaulted to a filing jurisdiction early in the EU implementation process should take a fresh look. Regulatory guidance has matured significantly, and there may be a more efficient or better-suited path available than the one originally chosen.

The uncomfortable divergence

There is a notable irony in the current environment. Domestically, the IRS and U.S. Treasury’s 2025-2026 Priority Guidance Plan reflects an explicit focus on deregulation and burden reduction, detailing dozens of projects aimed at reducing compliance costs for US businesses. Meanwhile, the international compliance environment has moved in the opposite direction, adding disclosure layers, explanatory requirements, and public transparency obligations that many US businesses cannot avoid simply because they are headquartered in the United States.

This divergence has a direct implication for how tax departments allocate resources and make the internal case for investment in international compliance infrastructure. The burden internationally is not going down — indeed, it is intensifying — and that argument is now backed by concrete examples rather than projections.

3 things worth doing now

There are several actions that corporate tax teams should consider, including:

Audit CbCR data quality with Australia’s enhanced requirements in mind — If you cannot readily reconcile taxes accrued to taxes credited at the jurisdictional level, that gap needs to be closed before it becomes an authority inquiry.

Revisit EU filing jurisdiction strategy — If your jurisdictional decision was made at the time of initial implementation and has not been reviewed since, it is worth a fresh look before the next reporting cycle.

Develop an internal narrative around public CbCR data before it circulates externally — Your company’s tax story should not be a surprise to the corporate teams involved in communications, investor relations, or ESG — and in today’s world, assuming such news stays quiet is no longer a safe assumption.

While CbCR started as a tool for tax authorities, it today has become something more visible, more public, and more consequential than that — and that trajectory is not reversing any time soon.

You can download a full copy of the Thomson Reuters Institute’s

Key takeaways:

• • • Agentic AI poised for adoption uptick — Agentic AI is following GenAI’s rapid adoption in the legal industry, with less than 20% of firms currently implementing agentic systems but half planning or considering adoption in the near future, according to a new report.

    • Adoption depends on human oversight answers — Legal professionals are generally optimistic about agentic AI’s potential, but successful adoption depends on explicit guidance about human oversight and the lawyer’s role in maintaining ethical standards.

    • Time to retool AI education? — Agentic AI’s increased autonomy introduces new oversight and ethical challenges for law firms, making targeted education and clear guidance essential to understanding the differences from GenAI.

Over the past several years, law firms and corporate legal departments have turned towards generative AI en masse. At the beginning of 2024, just 14% of all law firms and legal departments featured an enterprise-wide GenAI tool. Just two years later, that number had already risen to 43% of all firms and departments, according to the 2026 AI in Professional Services Report, from the Thomson Reuters Institute (TRI). For large law firms or legal departments, those percentages — not surprisingly — are beginning to approach 100%.

With GenAI adoption now this widespread, legal industry leaders are now turning their attention to two primary initiatives. One, of course, is how to get the most out of the AI tools they already have — a task that is proving a bit elusive. Currently, less than 20% of lawyers say their organizations measure AI’s return-on-investment, and most corporate lawyers say they have no idea how their outside law firms are approaching AI. Thus, instituting not just AI tools, but also an AI strategy is the second top priority for law firms and corporate legal departments in 2026 and beyond.

However, even as the legal industry reaches a tipping point in adopting GenAI tools, technology innovation still continues unabated. Agentic AI has emerged as the next wave of innovation that could change how lawyers work on a daily basis, offering a way to autonomously complete multi-step tasks. For example, agentic AI systems are already being built for the legal industry that independently researches a regulation or law, drafts a document based on the finding, identifies pitfalls, and revises the document, with stops for human guidance only instituted as desired.

According to the AI in Professional Services Report, the legal industry is already making headway towards implementing agentic AI systems. For agentic AI to truly take hold in legal, however, lawyers still require more education around not only how it differs from the GenAI systems they already have in place, but also when and where human intervention needs to occur within an agentic system.

The early stages of agentic AI

Examining current agentic AI adoption for the legal industry almost takes one back in time — two years, to be exact. Following the public release of GenAI in late-2022, many legal industry organizations spent 2023 evaluating and experimenting with AI systems, usually with a small working group of interested guinea pigs. As a result, only 14% of survey respondents said their law firms or corporate legal departments were engaged in organization-wide GenAI rollouts at the start of 2024. However, more than half of respondents said their organizations expected to be rolling out large-scale GenAI systems over the next 1 to 3 years. The intervening two years since then have proved that prediction to be largely true.

Agentic AI usage in the first half of 2026 looks largely similar to GenAI in 2024. The legal industry started to experiment with agentic AI at the beginning of 2025, with an eye towards actual implementation in 2026 and beyond (particularly as legal software providers began to integrate agentic systems into their own products). As such, less than 20% of recent survey respondents say their organization is engaged in widespread agentic AI adoption, but with about half of respondents said their organization is either planning to use or considering whether to use agentic AI in the near future.

By and large, lawyers feel positive about the agentic AI movement. When asked about their sentiment towards agentic AI, 51% of legal industry respondents said they felt excited or hopeful, while just 19% said they felt concerned or fearful. Further, about half (47%) said they actively believe agentic AI should be used for legal work, while 22% felt it should not, with the remainder saying they were unsure. These figures largely track with the sentiments expressed about GenAI in 2024, which have only grown over time from about 50% positive two years ago to two-thirds of all legal professionals feeling positive currently.

This all lends further credence to a rise in agentic AI usage similar to what law firms and corporate legal departments experienced with GenAI over the course of 2024 and 2025. Indeed, when asked when they expect agentic AI to be a central part of their workflow, few have baked agentic systems into their daily work currently, but a majority of legal industry respondents expect it to be central within the next 3 to 5 years.

The unique barriers of agentic AI adoption

Agentic AI does differ from GenAI in one crucial area that may limit its growth potential within the legal industry, however — autonomy. By and large, GenAI systems operate on a back-and-forth basis: Users provide the tool a prompt, receive its output, and then iterate back-and-forth from there. Agentic AI is intended to be more automated by design, only requiring human input at pre-determined points in the process. And that makes some lawyers understandably nervous.

When asked why they might feel hesitant about using agentic AI for legal tasks, the most common answer was a general fear of the unknown, but the second most common answer dealt with the need for careful monitoring and oversight. In fact, some respondents said they were excited about GenAI, but more cautious about agentic AI’s potential.

“Agentic AI, while exciting, to me removes oversight a step too far,” said one such lawyer from a US law firm. “I like the idea of prompting and reviewing a result. It is something else to have a machine have so much autonomy in the actual doing of a thing and potentially acting on my behalf without that very concrete review.”

Agentic AI usage in the first half of 2026 looks largely similar to GenAI in 2024.

An assistant GC at a US company also pointed to potential privacy and security concerns, adding: “The fact that agentic AI operates in a much more autonomous way, with a lack of control from the user, means there are many unknowns that are hidden beneath the process.”

For law firm and corporate legal department leaders looking to potentially implement agentic AI systems into their practice, this means re-thinking what AI education and training will mean moving forward. Beyond that, however, legal AI educators also will need to make sure to pinpoint and perhaps over-explain those specific instances in which human oversight needs to occur in agentic systems. More autonomous does not mean fully autonomous, and particularly for lawyers with ethical duties to their work product, lawyer oversight will in fact be a necessary part of any agentic system.

For law firm or legal department leaders, that means that finding the right balance between efficient workflows and human intervention will be key to agentic AI adoption. And those organizations that can best communicate human-in-the-loop to their professionals up-front will be rewarded with more increased and reliable adoption.

Clearly, lawyers feel positively about the agentic AI future, after all. They just need it spelled out explicitly as to what the lawyer’s role will be in this new paradigm.

“Agentic AI is powerful, but its moral compass must come from humans,” one UK law firm barrister noted aptly. “Lawyers are trained to safeguard fairness, rights, and the rule of law — principles that should guide how AI is designed, governed, and deployed. Hope lies in our ability to shape AI through these values for fairer values for society as a whole.”

You can download a full copy of the Thomson Reuters Institute’sĚý2026 AI in Professional Services ReportĚýhere

Key takeaways:

• • • Only IEEPA‑based tariffs are up for refund — Refunds will flow electronically to importers of record through ACE, the government’s digital import/export system, but only once CBP’s process is finalized.

    • Liquidation and protest timelines are now critical — An organization’s tax concepts that directly influence which entries are eligible and how long companies have to protect claims.

    • Tax functions must quickly coordinate with other corporate functions — In-house tax teams need to coordinate with their organization’s trade, procurement, and accounting functions to gather data, assert entitlement, and get the financial reporting right on any tariff refunds.

WASHINGTON, DC — When the United States Supreme Court issued its much-anticipated ruling on President Donald J. Trump’s authority to impose mass tariffs under the International Emergency Economic Powers Act (IEEPA) in February it set the stage for what it to come.

The Court ruled the president did not have authority under IEEPA to impose the tariffs that generated an estimated $163 billion of revenue in 2025. In response, the Court of International Trade (CIT) issued a ruling in requiring the U.S. Customs and Border Protection (CBP) to issue refunds on IEEPA duties for entries that have not gone final. That order, however, is currently suspended while CBP designs the refund process and the government considers an appeal.

AtĚýthe recent , tax experts discussed what this ruling means for corporate tax departments, outline what is and isn’t a consideration for refunds and the steps necessary to apply for refunds.

As panelists explained, the key issue for tax departments is that only IEEPA tariffs are in scope for refund — many other tariffs remain firmly in place. For example, on steel, aluminum, and copper; Section 301 tariffs on certain Chinese-origin goods; and new of 10% to 15% on most imports still apply and will continue to shape effective duty rates and supply chain costs.

So, which entities can actually get their money back?

Legally, CBP will send refunds only to the importer of record, and only electronically through the government’s digital import/export system, known as the Automated Commercial Environment (ACE) system. That means every potential claimant needs an with current bank information on file. And creating an account or updating it can be a lengthy process, especially inside a large organization.

If a business was not the importer of record but had tariffs contractually passed through to it — for example, by explicit tariff clauses, amended purchase orders, or separate line items on invoices — they may still have a commercial basis to recover their share from the importer. In practice, that means corporate tax teams should sit down with both the organization’s procurement experts and its largest suppliers to identify tariff‑sharing arrangements and understand what actions those importers are planning to take.

Why liquidation suddenly matters to tax leaders

As said, the Atmus ruling is limited to entries that are not final, which hinges on the . CBP typically has one year to review an entry and liquidate it (often around 314 days for formal entries) with some informal entries liquidating much sooner.

Once an entry liquidates, the 180‑day protest clock starts. Within that window, the importer of record can challenge CBP’s decision, and those protested entries may remain in play for IEEPA refunds. There is also a 90‑day window in which CBP can reliquidate on its own initiative, raising questions about whether final should be read as 90 days or 180 days — clearly, an issue that will matter a lot if your company is near those deadlines.

Data, controversy risk & financial reporting

The role of in-house tax departments in the process of getting refunds requires, for starters, giving departments access to entry‑level data showing which imports bore IEEPA tariffs between February 1, 2025, and February 28, 2026. If a business does not already have robust trade reporting, the first step is to confirm whether the business has made payments to CBP; and, if so, to work with the company’s supply chain or trade compliance teams to access ACE and run detailed entry reports for that period.

Summary entries and heavily aggregated data will be a challenge because CBP has indicated that refund claims will require a declaration in the ACE system that lists specific entries and associated IEEPA duties. Expect controversy pressure: As claims scale up, CBP resources and the courts could see backlogs. If that becomes the case, tax teams should be prepared for protests, documentation requests, and potential litigation over entitlement and timing.

On the financial reporting side, whether and when to recognize a refund depends on the strength of the legal claim and the status of the proceedings. If tariffs were listed as expenses as they were incurred, successful refunds may give rise to income recognition. In cases in which tariffs were capitalized into fixed assets, however, the accounting analysis becomes more nuanced and may implicate asset basis, depreciation, and potentially transfer pricing positions.

Coordination between an organization’s financial reporting, tax accounting, and transfer pricing specialists is critical in order that customs values, income tax treatment, and any refund‑related credits remain consistent.

Action items for corporate tax departments

Corporate tax teams do not need to become customs experts overnight, but they do need to lead a coordinated response. Practically, that means they should:

• • • confirm whether their company was an importer of record and, if so, ensure ACE access and banking information are in place now, not after CBP turns the refund system on.
    • map which entries included IEEPA tariffs, identify which are non‑liquidated or still within the 180‑day protest window, and file protests where appropriate to protect the company’s rights.
    • inventory all tariff‑sharing arrangements with suppliers, assess contractual entitlement to pass‑through refunds, and align with procurement and legal teams on a consistent recovery approach.
    • work with accounting to determine the financial statement treatment of potential refunds, including whether and when to recognize contingent assets or income and any knock‑on effects for transfer pricing and valuation.

If tax departments wait for complete certainty from the courts before acting, many entries may go final and fall out of scope. The opportunity for tariff refunds will favor companies that are data‑ready, cross‑functionally aligned, and willing to move under time pressure.

You can find out more about the changing tariff situation here