AI in accounting is moving into a new phase. It's no longer just about helping people complete individual tasks faster. It's about connecting AI to the clients, workflows and knowledge of the practice itself. For accounting firms, that opens up new possibilities. But the real opportunity isn't simply having access to more AI tools. It's building a practice that knows how to use them effectively.
Many accountants are already using AI to draft emails, summarise documents and research questions. These are useful starting points, but they often require users to provide the context and information needed for each task.
The next step is AI that works within the systems firms already use, with access to relevant client information, communications and workflows.
This is the direction Karbon is taking with its AI-powered practice management platform and Kai, its AI built into Karbon.
Rather than starting from scratch with every request, Kai understands the context of a firm's clients, work, team and data. This creates opportunities to ask firm-specific questions, surface priorities and bring together information that would otherwise require manual searching.
Imagine starting your day by asking which client jobs need attention, what is holding up a month-end close or which outstanding requests could delay work. Instead of searching through emails, work items and client records, AI can help bring the relevant context together.
Karbon has also introduced AI agent capabilities that recommends and actions next steps based on review of a firm’s data. These AI Agent workflows in Karbon can surface what needs attention based on unbilled WIP, client health, scope creep, outstanding client requests, and colleague performance. Customers can customize any of them or build its own. Karbon's wider AI capabilities also support everyday tasks such as summarising email conversations, drafting client replies and suggesting work assignments.
The aim is to spend less time finding information and more time acting on it. This points towards a future where AI becomes more closely connected to how work moves through a practice.
The real opportunity goes beyond productivity
Saving time is valuable, but it doesn't automatically create a better business.
If a firm uses AI to complete the same work more quickly but doesn't reconsider how it operates, it may miss a much bigger opportunity.
Time saved on repetitive administration could create capacity for better client conversations, more proactive advice and new services. Better access to information could help teams identify problems earlier, understand client needs and make more informed decisions.
But these outcomes don't happen simply because a firm introduces AI.
They depend on the quality of the underlying information, how well AI understands the task, the boundaries within which it operates and the people responsible for using it.
Five pillars of AI success
In our joint App Advisory Plus and Karbon webinar, The Five Pillars of AI Success: Building Your Practice's Operating System, we explored the foundations firms need to make AI work effectively.
The five pillars — Input, Context, Guardrails, QA and Enablement — provide a framework for moving beyond isolated experiments.
1. Input: Start with reliable information
AI can only work with the information available to it. Incomplete records, inconsistent processes and poor-quality data can undermine the results, however capable the technology.
2. Context: Give AI the bigger picture
A useful answer depends on understanding the client, the task and the circumstances. Connecting AI to relevant firm information can make its output more useful than a generic response without that context.
3. Guardrails: Define what AI can do
Firms need clear boundaries around access, permissions and actions. They should understand which tasks AI can support, when approval is required and where decisions must remain with an authorised person.
4. Quality assurance: Check the result
AI output shouldn't automatically be treated as correct. Firms need proportionate review processes, clear accountability and a way to identify and address errors.
5. Enablement: Prepare your people
Technology alone won't transform a practice. Teams need training, confidence and clear guidance on when to use AI, when to question its output and how to apply their professional judgement.
These pillars work together. A capable AI tool will struggle to deliver consistent value if the information is unreliable, the process is unclear or the team doesn't know how to use it.
As AI develops from generating suggestions towards carrying out tasks, questions of trust and accountability become increasingly important.
Who is responsible for an action taken with AI assistance? What should happen when an output looks wrong? How can a firm demonstrate that sensitive information has been handled appropriately?
These aren't reasons to avoid AI. They're questions firms need to answer as part of implementation.
Karbon's approach places AI within its practice management environment, where relevant client and workflow context is already held. And crucially, human approval steps and controls around data access and the ability for firms to manage AI functionality have been designed throughout, ensuring that security and data privacy are maintained.
Whatever platform a firm chooses, it should understand what information its AI can access, how its outputs are checked and which actions require human approval. Professional judgement and accountability remain essential.
You don't need to transform your whole practice overnight. Start with a problem worth solving.
Where is your team losing time?
Look for repetitive tasks, manual information gathering, unnecessary handovers or processes that regularly hold up client work.
What outcome do you want to improve?
Be specific. Are you trying to reduce turnaround times, improve visibility of work, strengthen client communication or create more capacity for advisory services?
How will you keep control and measure success?
Decide what information the AI needs, who will review its output and how you'll know whether the change has worked. Test the approach on a manageable use case, gather feedback and refine it before expanding further.
AI adoption isn't simply a technology project. It's a question of how a firm organises its work, supports its people and maintains the quality its clients expect.
The practices that benefit most will be those that combine useful technology with reliable information, well-designed processes and human expertise.
The question isn't just which AI tool to try next. It's whether your practice is ready to put AI to work.
Explore further
Watch our joint webinar: The Five Pillars of AI Success: Building Your Practice's Operating System
Explore the research: The State of AI in Accounting Report 2026
Read more: The 3 waves of AI impact, and how your firm creates value in each
Latest news, events, and updates on all things App related, plus useful advice on App advisory - so you know you are ahead of the game.
Connect with us
