TAINA Technology Network_Shape_Transparent.png

Agentic AI in Tax Operations

Artificial intelligence is advancing at an extraordinary pace.

For the first time, we are seeing technologies capable of

transforming how complex operational work is performed across financial institutions.

In tax operations, the opportunity is particularly significant.

Highly skilled professionals across financial institutions still spend large portions of their time reviewing documentation, validating forms, identifying inconsistencies, and following up on missing information.

These tasks are critical for compliance, but they are also repetitive and time-consuming.

Agentic AI creates the opportunity to rethink how these processes operate.

By reasoning through workflows, validating data and identifying issues earlier in the process, intelligent systems can significantly reduce manual remediation work while improving accuracy and consistency.

The real opportunity is not replacing people, but allowing professionals to focus where human judgement adds the most value, oversight, risk management and strategic decision-making.

For financial institutions facing growing regulatory complexity and operational costs, the potential return on investment is significant.

Below are a short series of reflections on how AI may reshape tax operations, the opportunities for real value creation, the risks institutions must manage, the lessons we are learning as we build these technologies, and the potential impact on the talented professionals working across our industry.

When AI Acts on Its Own: A Real‑World Lesson in Agentic Governance

Following on from my previous post on governance, risk, and control in AI, I wanted to share a real-life example that happened to me this week, which served as a very real reminder of why this matters.

As we move into the world of Agentic AI, we are not just building systems; we are introducing actors into our environments. With that comes new and sometimes unexpected behaviours.

Recently, we observed something unusual. An unsupervised AI agent attempted to join one of our internal investor update meetings to take notes. What made this more interesting was that it had not been instructed to do so by its “manager”.

On one level, this is exactly what makes Agentic AI so powerful. But it is also exactly what makes it risky.

These systems do not simply follow instructions. They interpret, act, and optimise. And sometimes, they overstep. For us, this was a valuable moment. Not a failure, but a signal. A reminder that as we build these capabilities, we must remain vigilant.

In response, we have implemented additional controls to ensure that AI agents cannot access meetings or sensitive environments without explicit permission. This required both technical measures and governance controls, reinforcing that oversight must remain firmly in human hands.

And this is just one example. The reality is that we will continue to face new situations like this, situations we have not seen before, and decisions we have not had to make before. The challenge is not just building AI. It is managing behaviour in dynamic environments.

The firms that succeed will not be those who avoid these moments, but those who recognise them early, respond thoughtfully, and continuously strengthen their governance frameworks. Because in the age of Agentic AI, control is not a one-time design. It is an ongoing discipline.

How Agentic AI will deliver tangible ROI to Tax Operations

Up until now, automation in tax had to be done in multiple individual steps – e.g. validate a form, reconcile data for reporting, submit tax report to the tax authorities. And yet the real inefficiency and operational risk and cost sits in the gaps between these steps as they still have to be done with a lot of human manual intervention. Agentic AI changes that.

Instead of automating tasks in isolation, it will enable end-to-end workflows, connecting processes, data, and decisions across the full lifecycle.

This is where the real ROI emerges:

Reduced operational risk
Fewer manual interventions and reconciliations
Cleaner, more reliable data
And teams freed up to focus on higher-value, higher-risk work

A few simple examples:

A TIN validation agent verifies the TIN
Quality Assurance Agent checks verification is accurate
If accurate, Admin Agent updates the trusted customer record.
A Compliance Agent monitors new regulatory guidance and triggers alerts with recommended next steps when changes are identified.
A form expiry agent identifies upcoming deadlines and triggers an alert, initiates re-solicitation workflows automatically.

Individually, these are useful. Connected together, they transform the operating model.

Workflows can be configured to reflect how teams actually operate, powered by multiple specialised agents working in sequence, with built-in quality controls at every stage.

This is a step change from today’s fragmented processes, where even with automation, coordination remains manual.

Agentic AI doesn’t just make tax operations more efficient. It empowers them to be a true competitive lever.

We’re starting to see this shift in how leading institutions are thinking about tax operations. I’d be interested to hear, how do you think about the place of AI in Tax Operations?

The Real ROI of Agentic AI: Redesigning How Work Gets Done

The real ROI of Agentic AI in tax operations isn’t just productivity — it’s how work itself will change.

We are on the cusp of a fundamental shift in organisational design — one that I believe will unlock faster innovation, higher productivity, and far greater fulfilment at work.

Agentic AI enables teams to create, test and iterate hundreds of times faster than ever before. In doing so, it challenges traditional models of “pure” subject‑matter expertise that sit outside real‑world application. Going forward, expertise must live inside the process — embedded in rapid cycles of creation, learning and improvement, powered by proactive, multidisciplinary collaboration.

This means smaller, dynamic teams made up of diverse skills and perspectives, working together in agile formations. While we often talk about this in the context of product development, the same principles apply just as much to traditionally conservative areas of financial institutions — from goal setting and reporting to audits and compliance.

For leaders, the role evolves too: assembling the right mix of talent at the right time, setting clear objectives, and ensuring strong quality assurance from deployment through to continuous improvement. The payoff is clear — significantly greater agility, speed of delivery, and economic ROI.

But just as importantly, this model creates work that is richer and more meaningful for individuals. Professionals can apply their skills across a wider range of challenges, see the impact of their contribution, and grow through variety rather than repetition. These structures naturally attract more diverse talent and foster more inclusive cultures — a benefit for organisations and people alike.

At TAINA, we’ve already begun pioneering this approach. We call them Mission Teams. Within our Agentic AI for Tax Operations mission, developers, engineers, tax experts, designers, customer success, and marketing work together — flexing as the product evolves. The early results are incredibly promising, and I’m genuinely excited about what this way of working makes possible.

The future of work in tax isn’t just automated.
It’s adaptive, collaborative — and deeply human.

Building AI Capability Without Compromising Trust

As AI adoption accelerates across financial services and the rapid rise of unvetted, easily accessible AI tools, one question is becoming increasingly important: Are we building capability… or introducing risk?

There is a growing temptation to experiment with unvetted models and non-enterprise solutions. These tools are powerful, accessible, and fast to deploy. However, in regulated environments, speed without control is not innovation; it is exposure.

Financial institutions are not simply working with data. They are custodians of highly sensitive client information and operate within strict regulatory frameworks. This fundamentally changes how AI must be approached. AI cannot be treated as a standalone tool. It must be introduced as part of the operating model, with the same level of rigor applied to any core system.

This means ensuring:
• Privacy is designed into the system from the outset
• Clear governance frameworks define how AI is used, trained, and monitored
• Outputs are auditable, and decisions can be explained
• Data access, movement, and retention are tightly controlled
• Only enterprise-grade, production-ready solutions are deployed

The journey from experimentation to production is significant and cannot be rushed. In financial institutions, safety, reliability, and governance are non-negotiable.

At TAINA, this approach is underpinned by ISO27001-aligned controls, strict software governance, and continuous monitoring of AI usage as part of an evolving programme that adapts to emerging threats.

The reality is simple:
👉 Not all AI is created equal
👉 And not all AI is suitable for regulated environments

The organisations that will lead are not those who move fastest, but those who build secure, governed, and scalable AI foundations. Because in this space, trust is everything. And trust is built on control.

From Automation to Advantage: The Evolution of Tax Operations in the Age of AI

As AI begins to automate some of these operational activities, the role of the tax professional will likely evolve.

Historically, many tax operations teams have relied heavily on manual validation and remediation processes. While essential, these workflows often leave highly capable professionals spending much of their time on repetitive tasks.

To thrive in this shift, we must empower every team member to understand and embrace the new reality. The pace at which each of us must execute must accelerate 100x. This means everything that can be automated with tools must be, as a matter of urgency, and each of us has to lean into our uniquely human ROI generators.

We may begin to see greater emphasis on:
• Data analysis
• Technology fluency
• Critical thinking
• Cross-functional collaboration

Rather than simply processing documentation, tax teams will increasingly oversee automated systems, interpret insights and manage complex regulatory environments.

In that sense, AI may actually make tax operations a far more attractive and intellectually engaging field for the next generation of professionals.

Preserving Institutional Knowledge

One of the less discussed challenges in tax operations is the preservation of institutional knowledge.

Many financial institutions rely heavily on experienced tax professionals who hold years of operational expertise, knowledge that is often difficult to document or replicate.

When experienced team members move on, a significant amount of this knowledge can be lost.

AI systems have the potential to help capture patterns, decision frameworks and operational insights over time.

Over time, they can begin to serve as a form of organisational memory, making institutions more resilient and less dependent on fragmented or individual knowledge.

When implemented responsibly, this could allow institutions to preserve expertise while maintaining greater consistency across teams and jurisdictions. In a regulatory environment that continues to grow more complex, that kind of institutional memory may become a critical source of resilience for financial institutions.

AI CAN HELP INSTITUTIONS PRESERVE OPERATIONAL EXPERTISE (11).png

Moving from Remediation to Intelligence

AI discussions often focus heavily on efficiency gains.

But in tax operations, the impact may be broader than simply doing things faster.

As regulatory frameworks such as FATCA, CRS and emerging developments like CRS 2.0 continue to expand, the operational demands placed on financial institutions are increasing significantly.

Tax operations teams are being asked to manage growing volumes of documentation, validation requirements and remediation workflows.

AI offers an opportunity to rethink how these processes operate.

By introducing intelligent validation, real-time data checks and earlier detection of issues, institutions may be able to move from reactive remediation toward more proactive compliance models.

The long-term impact could be significant, not only improving efficiency, but reshaping how tax operations functions support the broader organisation.

AI CAN HELP INSTITUTIONS PRESERVE OPERATIONAL EXPERTISE (4).png

Leadership

The pace of technological change is accelerating, and with it comes a new responsibility for leadership.

Every organisation must consider how it supports its teams in adapting to this shift.

This means creating an environment where new technologies are understood, embraced and applied thoughtfully.

It also means recognising where automation can deliver value, and where human expertise remains essential.

The real opportunity lies in enabling people to focus on areas where human judgement, creativity and strategic thinking make the greatest impact.

Organisations that take a proactive approach will not only improve operational performance, but will also empower their teams to operate at a higher level.

In this new era, leadership is not just about adopting technology, it is about guiding people through change.