CFOAgent · AI Implementation
AI Consulting · Accounting & Controlling · SwitzerlandImplementing AI in finance and accounting.
AI consulting from practice: 20 years in finance and accounting as CFO and interim manager. We implement AI directly through concrete use cases, e.g. in payables, general ledger, controlling and consolidation. We work with your staff so that your team can develop the use cases further on its own. Already delivered in numerous SMEs and corporates.
How an AI implementation runs with us.
We do not start with a weeks-long analysis. We start in day-to-day operations. Together with the people in charge, we look at the actual workflows in bookkeeping, controlling and reporting: how documents come in, how the monthly close is built, where things get reconciled, commented, copied and rebuilt.
Because we have seen a great many companies with completely different accounting and controlling processes, across industries, system landscapes and company sizes, we quickly recognise which processes are suited to artificial intelligence and which are not. The patterns repeat, the details differ everywhere. Exactly that combination makes prioritisation fast.
After a short time, a concrete implementation plan is on the table: Which use cases deliver the most value? Which tools fit? Which systems get connected? Which projects come first, and what do we deliberately leave alone?
Our strength is not knowing as many AI tools as possible. Our strength is recognising very quickly, based on decades of work in finance and accounting, where AI creates real value in your specific environment, and where it does not. That applies to AI in finance as a whole: from the general ledger to controlling to management reporting.
- No lengthy presentations and no theoretical AI strategy papers.
- No months of as-is documentation before anything happens.
- No tool zoo: tools follow the use cases, not the other way round.
- No licence sales: we earn nothing on other vendors‘ licences, we recommend what fits.
From day-to-day operations to a plan, from the plan to implementation.
Review the workflows
On site, in daily business, with the people who do the work. Not interviews about processes, but the processes themselves.
Identify use cases
Where is repetitive work, where are media breaks, where is Excel acrobatics. Every use case is rated by value and feasibility.
Build the plan
A concrete roadmap: which projects first, which tools, which systems, who does what. Pragmatic rather than exhaustive.
Run pilots
The first use cases go into live operation, together with your staff, at their workplaces.
Anchor it
Training, ground rules and further development: AI becomes part of daily work, not a project that fizzles out.
Where we implement AI in accounting.
Every company is different, but these are the use cases we have implemented again and again in SMEs and corporates:
Payables
From inbox to posting proposal: capturing, coding and checking recurring documents.
General ledger & reconciliations
Bank, clearing accounts, accruals: finding, documenting and preparing differences instead of hunting for them.
Monthly close
Checklists, plausibility checks on the income statement, spotting anomalies before someone finds them in the report.
Reporting & commentary
Compiling reports and commenting variances cleanly, in your language and your format.
Consolidation
Reconciling intercompany, mapping the entities‘ charts of accounts, checking and documenting consolidation entries.
Controlling & ad-hoc analysis
Answering questions about the numbers in minutes instead of days, down to the individual document.
Planning & forecast
Data preparation, driver logic and scenarios: less copy-paste, more time for the message.
Excel automation
Defusing the grown Excel chains: automate what eats time daily, replace what is dangerous.
Whether the answer is an AI agent, a lean automation or deliberately nothing at all is decided case by case. Sometimes the best recommendation is: simplify this process first, then automate it.
20 years of practice, more than 80 mandates in finance and accounting.
Behind CFOAgent’s AI consulting stands Roman Kalberer: 20 years in finance and accounting as CFO and interim manager, more than 80 mandates as CFO, head of accounting, controller and project lead, in corporates, SMEs and the public sector. Plus more than 10 years of AI, BI and ERP practice, and a lecturer on AI in finance and accounting. That shapes how we implement AI.
Since the launch of ChatGPT we have been training professionals in the use of AI: as a lecturer at Swiss education institutes such as SwissAccounting and the Controller Akademie, and in workshops directly in numerous SMEs. Our key lesson: teams learn most not in the classroom, but by working on their own use cases. That is exactly how we implement AI today:
- We speak accounting: account assignment, VAT, subledgers, reconciliations and closing processes are our daily business, not foreign words.
- We recognise patterns fast: whoever has seen many finance organisations sees in hours what takes others weeks.
- We work with your people: implementation happens at the workplace, on your staff’s own use cases, not in the meeting room. Your team then develops the solutions further on its own, without depending on us.
- We stay independent: recommendations follow value, not licence models. From the Microsoft stack to specialised tools.
For teams that answer for the numbers.
Accounting and controlling departments
The finance department with several employees, several entities, grown processes and reporting chains: this is usually where the greatest potential lies, and where a partner counts who knows group processes from his own responsibility.
Finance teams with several employees
A well-rehearsed finance team with grown workflows: AI creates room for analysis and steering, without needing an IT department for it.
With all common AI tools. With a focus on Claude Code and Codex.
We work with the AI tools that are available and permitted in your company, in SMEs just as in corporates. We have a strong focus on Claude Code and Codex: they produce analyses and automations in a short time that work directly with your data and stay traceable. Where data must not leave the building, we deliberately use local models.
Claude Code and Codex
AI that does not just write text, but processes files, builds analyses and automates workflows. Strong in reconciliations, Excel chains, reporting and data preparation.
What you already have
Microsoft Copilot, ChatGPT, Gemini or the AI features in Abacus, Bexio, Topal, Odoo and Business Central: we get meaningful results out of the licences you already have.
Local AI models
When confidential financial data must not go into a cloud, we deliberately use local models. They run on your own infrastructure, the data stays in-house.
The CFOAgent booking agent
A large part of an AI implementation can be a booking agent: it receives documents, codes them with reasons and posts after approval. It can be an important component, but it does not have to be. The implementation shows whether it fits for you.
AI implementation, answered briefly.
How does an AI implementation in accounting start sensibly?
Not with a strategy presentation, but with the actual workflows: documents, closing, reconciliations, reporting. The review produces the use cases, and the use cases produce the plan. That way you invest first where the value is greatest.
How quickly is there a concrete plan?
Considerably faster than in classic consulting projects: the review takes days, not months. Because we know a great many finance organisations, prioritisation is fast, and the plan contains concrete projects, tools and responsibilities instead of declarations of intent.
Which tools do you use?
The ones that fit your environment. It often starts with the tools you already licence, such as the Microsoft stack or the AI features of your ERP. We have a strong focus on Claude Code and Codex. Where it pays off, our CFOAgent booking agent comes in as an optional component. We earn nothing on other vendors‘ licences.
Do our staff need prior AI knowledge?
No. Implementation happens at the workplace, with your staff’s real cases. Whoever knows their process learns to use AI within it surprisingly fast. Training and ground rules are part of the implementation.
What about data protection and confidentiality?
We clarify that per use case: what may go to which cloud, what stays in-house, what anonymisation is needed. There is a viable path for practically every requirement, from Swiss cloud hosting to local models. The important thing is to ask the question at the start, not at the end.
Can our team develop the use cases further on its own?
Yes, that is the goal. We implement the first use cases together with your staff, on their own cases. Along the way they learn how to identify, build and check use cases. Your team develops the next use cases on its own; we only accompany you if you wish.
Does AI replace our accountants and controllers?
No, it changes their work: less capturing and retyping, more reviewing, steering and analysing. At a time when finance professionals are hard to find, that is relief rather than threat.
What does an AI implementation cost?
That depends on scope. The intro call is free. The entry via the first use cases is deliberately lean, so value and cost become visible early. After the intro call you receive a clear estimate.
What sets you apart from classic AI consulting?
We come from finance and accounting, with 20 years of practice as CFO and interim manager, not from technology consulting. Instead of concepts and roadmaps we deliver running use cases in daily operations, and we work with your staff until they develop them further on their own.
Does this work in our industry?
Very likely: we have supported finance processes in manufacturing, construction, trade, services and the public sector. The basic accounting patterns are similar across industries; we pick up your industry specifics in the review.
Four companies, clear roles.
Whenever a project needs more than AI implementation, the paths are short: data platform, interim reinforcement or sparring for leadership come from the same group.
Let’s talk about your processes.
In the intro call we listen, ask the right questions and tell you honestly where we see potential and where we do not. Then you decide whether a review makes sense.
Book an intro call