WHITE PAPER
Beyond omnichannel: How AI is reshaping indirect tax compliance
The compliance environment has changed — the model has not
The tools most professionals rely on were built for a different era. AI closes that gap — but only when it is built to the standard the work demands.
Ask an indirect tax professional where they want to spend their time, and the answer is consistent. The 2025 State of the Corporate Tax Department from the Thomson Reuters Institute found that tax professionals currently spend more than half their working hours on reactive, tactical compliance work. What they want is nearly the opposite: 70% of their time on proactive, strategic activity. That gap is structural, not motivational.
The compliance environment these professionals are working in today is fundamentally different from the one automation was built for. U.S. sales and use tax spans more than 11,000 distinct filing jurisdictions, and tax content coverage across all taxing authorities — including county and city sub-jurisdictions — extends to more than 19,000. Rates and rules change continuously. Product taxability in retail is unusually complex, and it shifts with legislative and regulatory updates in ways that generic rule sets struggle to keep up with.
Marketplace facilitator obligations and buy-anywhere/return-anywhere models create data-reconciliation problems that sit upstream of every filing. A “buy online, pickup in-store” (BOPIS) transaction is a single taxable event sourced to where the customer takes possession, but the sale originates in one system and the fulfillment records in another, and applying the correct sourcing logic requires those two data streams to be consistent from the moment the transaction occurs. When they are not, the error does not surface until the return is being prepared.
Then there is peak trading. Black Friday and Cyber Monday alone can generate more exception volume in 72 hours than a typical compliance team processes in a month. The compliance infrastructure underpinning that volume needs to absorb it without expanding the team's manual workload, or the exceptions accumulate faster than they can be cleared.
A further structural pressure is emerging as AI agents begin initiating and completing commercial transactions autonomously — executing purchasing decisions, processing orders, and fulfilling obligations on behalf of businesses and consumers. The tax infrastructure underpinning those transactions needs to meet the same standard: real-time, accurate, and auditable at the point the transaction occurs. A manual compliance back end cannot support an agentic commerce front end, which means the compliance model has to be as autonomous as the transactions it governs.
Automation addressed volume and speed and was the right investment for its time. However, it optimized the existing model rather than changing it. Tax professionals are still bridging disconnected systems manually, resolving exceptions by hand, and making classification judgments their tools were never built to make. The question is not whether AI can help; it’s whether AI can be trusted to handle what compliance actually requires.
The takeaway
The compliance environment has changed structurally. Automation optimized the old model but did not replace it. AI changes what is possible — but only when it is designed to a standard the work can defend.
What AI actually does that automation could not
The majority (69%) of corporate tax departments are still operating in the chaotic or reactive stage, despite years of automation investment.*
Automation runs on explicit rules. It executes defined logic reliably and at scale, but it stops at the edge of ambiguity. When a product falls into a classification edge case, when a promotional mechanic changes the taxability of a bundled item, or when an exemption certificate is present but its scope is arguable — automation flags the exception and parks it. A human resolves it. That has always been the ceiling.
AI works in the space where rules run out. It applies judgment to classification questions that have no clean rule-based answer, drawing on jurisdictional guidance, legislative history, and authoritative tax content to reach a well-grounded conclusion. It resolves exceptions rather than surfacing them. It also learns from how prior exceptions were handled, so the queue of unresolved items shrinks over time rather than growing.
For omnichannel retailers, that difference is operationally significant. A BOPIS transaction is a single taxable event, but the sale and the fulfillment live in different systems, and sourcing logic needs to be applied correctly across both. When data from those two systems is inconsistent, the error does not get caught until the return stage. AI reconciles those data streams at the point of transaction, applies the correct sourcing logic, and moves on. Returns and exchanges generate offsetting obligations that rules-based systems handle inconsistently. Marketplace data arrives in different formats from different platforms. AI normalizes all of it, without a human at each handoff.
A promotional bundle — a taxable item and a food item sold at a single price — shows the difference clearly. Whether that bundle is taxable varies by state and the variation turns on rules that differ in kind, not just degree. A rules-based system applies whatever logic it was configured for and flags everything else. AI evaluates the bundle against the applicable guidance for each state, applies the correct treatment, and logs the reasoning. The professional does not see an exception queue; they see a completed return.
As professionals shift from handling every step to reviewing and approving finished output, the compliance cycle becomes less focused on execution and more judgment driven.
Touchless compliance: From execution to governance
Touchless compliance is a specific workflow shift, not a marketing term. The architecture runs autonomously from data ingestion through validation, mapping, and return preparation, delivering a review-ready output to the tax professional for final sign-off. The six manual stages of the compliance cycle do not disappear, but the manual intervention at each one does.
The tax professional's accountability does not diminish in this model; their exposure to manual execution does. When AI completes a compliance cycle, the professional receives a return where every data import has been reconciled, every exception resolved with a logged rationale, and every jurisdiction-specific requirement applied. If they want to examine a specific decision — why a product category was taxed at a given rate in a given state, why an exemption certificate was accepted or challenged, etc. — that reasoning is available within the return. They can review it, override it if their judgment differs, and file with confidence that the output is defensible. The audit trail is built into the return itself, not generated separately after the fact.
For omnichannel retailers managing hundreds of returns per period across dozens of states, that shift is the difference between a compliance function that is permanently reactive and one that has capacity for something else. Exception queues that once grew faster than teams could clear them are resolved before the professional sees the return. Reconciliation bottlenecks that pushed filing deadlines become a solved problem rather than a recurring one.
The takeaway
Touchless compliance eliminates the execution burden without transferring the accountability. The professional reviews, approves, and files — what changes is the state of the return when it reaches them.
Why trust is the real question, and what Fiduciary-Grade AI™ means
A significant portion (79%) of corporate tax professionals say agentic AI should be applied to their work, according to the 2026 AI in Professional Services Report. But appetite and trust are different things. Reliability and accuracy concerns remain the primary reason professionals hesitate. In sales and use tax, that hesitation is rational.
A return filed with incorrect rate application across a product category. An exemption certificate gap that surfaces in audit. A misclassified item generating back liability across multiple states. Compliance outputs carry professional accountability and the AI producing them needs to meet the same standard.
The Thomson Reuters standard for AI in high-stakes professional environments is Fiduciary-Grade AI. It defines the four conditions under which AI produces work that professionals can stand behind.
Those four principles become clearer when applied to an actual compliance decision. Consider a bundled transaction in a state that applies a food-component threshold test, where the question of whether the bundle is taxable turns on whether the food items represent more than a defined percentage of the total price.
The AI identifies the applicable jurisdictional rule, retrieves the authoritative guidance from the tax content library, calculates the component percentages, and applies the threshold. The rationale is logged in plain language within the return, with the rule cited, the calculation performed, and the outcome reached. If the professional's judgment differs — if they know the product categorization is under dispute or a state has recently issued administrative guidance that changes the analysis — they can override the decision. That override is recorded in the audit trail alongside the original AI reasoning. Nothing is hidden. Nothing requires a call to a support team to explain.
For U.S. retail tax teams, that standard is not a differentiating feature. It is the baseline requirement for AI that carries real compliance liability. A state revenue department conducting an audit does not take the filer's word for why a classification decision was made. With Fiduciary-Grade AI, the reasoning is in the return.
Requirements for AI compliance infrastructure
AI in tax compliance is expanding faster than most tax professionals have time to evaluate it. The concepts are consistent across vendors — agentic workflows, touchless processing, and human-in-the-loop review —but the capabilities behind them are not. For retail tax teams building compliance infrastructure that needs to absorb increasing complexity, the questions that matter most are operational, not conceptual.
The connected life cycle: Why point solutions create the next problem
The indirect tax life cycle begins at the point of transaction, not at filing. At the point of sale, tax determination calculates the correct rate and treatment, while in jurisdictions where e-invoicing is mandated, invoice validation occurs before the transaction completes. Compliance filing follows, reconciliation closes the cycle, and each stage flows into the next as part of a single workflow — not separate problems to be solved by separate tools.
For U.S. retailers with domestic operations only, fragmented point solutions create reconciliation failures and audit exposure at every handoff. Data flowing from determination through filing needs to be consistent; when it passes through separate tools with separate data models, it rarely is.
For retailers with international operations, the stakes are higher and the timeline is contracting.
International mandate examples
E-invoicing mandates are expanding globally at different paces. In Europe, France begins business-to-business (B2B) e-invoicing in September 2026, Germany follows in 2027, and the EU's ViDA cross-border B2B requirements take effect in 2030. In Latin America, clearance models are already live in multiple markets.
In clearance-model jurisdictions, an invoice that fails real-time validation is rejected before the transaction completes. That is not a filing error that surfaces in audit six months later — it stops the sale. In the U.S., retailers expanding internationally need compliance infrastructure connected across the full life cycle, not point solutions assembled after the fact that were never designed to work together.
From compliance processor to strategic advisor
More than half (58%) of corporate tax departments describe themselves as under-resourced, up from 51% the prior year.*
Most U.S. retail tax teams are small, but the compliance footprint they manage is not. High transaction volumes, a constantly shifting channel mix, and a jurisdictional landscape that never settles make the resource gap more acute in omnichannel retail than in almost any other sector.
The conventional responses — add headcount, increase outsourcing, ask existing staff to absorb more — do not scale as the compliance environment grows more complex and more continuous. They address symptoms rather than the model itself.
When AI manages data ingestion, validation, exception resolution, and return preparation, retail tax professionals get time back for the work their function was designed to do. They can assess the tax implications of entering a new state, model the impact of promotional pricing decisions, and contribute to commercial decisions with real fiscal consequences. Plus, they have time to build the kind of relationship with finance leadership that a team buried in execution rarely has capacity for.
That is not a future state. It is the practical consequence of removing six manual stages from a compliance cycle that repeats every filing period across thousands of jurisdictions.
The takeaway
The compliance function was never designed to spend most of its time on execution. Removing that burden does not change what tax professionals are accountable for — it changes what they spend their expertise on.
The window to act is closing
According to the 2026 AI in Professional Services Report, 77% of tax professionals expect agentic AI to be central to their compliance workflow by 2030. The obligations that U.S. omnichannel retailers are managing today are more complex, more continuous, and more consequential than they were five years ago. The retailers best positioned when those forces converge are not the ones waiting for the problem to become a crisis — they’re the ones building the infrastructure now.
For U.S. retailers, ONESOURCE Sales and Use Tax AI powered by CoCounsel delivers touchless compliance across more than 19,000 jurisdictions — the full compliance cycle automated from source data to signature-ready returns, with electronic filing in 33 states. Early customers report up to 65% reduction in time spent on routine reporting and a significant reduction in exceptions and errors reaching filing. It is software built to the Fiduciary-Grade AI standard.
For retailers with international operations, ONESOURCE Indirect Compliance AI powered by CoCounsel extends the same model globally, automating VAT return preparation, SAF-T, and digital filings across more than 60 countries, so that the compliance infrastructure built for U.S. complexity is the same infrastructure that absorbs international mandate requirements as they go live.
Evaluate your compliance infrastructure
The five criteria in this white paper provide a starting point for assessing whether your current compliance model is built for the complexity retail tax teams are navigating now, as well as the mandates arriving over the next four years.
To see what touchless compliance looks like in practice and to assess the criteria that matter for your operation, discover how ONESOURCE Sales and Use Tax AI works.
*Source: 2025 State of the Corporate Tax Department, Thomson Reuters Institute.
Sales and use tax software
Drive compliance forward with less stress
Learn how ONESOURCE Sales and Use Tax powered by CoCounsel helps your team replace manual tax tasks with AI-enabled workflows built for accuracy and control