Bogdan Năforniță, CEO and Co-Founder of Profluo, shared insights in an interview for the BusinessMark Blog, where he addressed the transformation of the finance department through AI technologies, the redefinition of how finance professionals work, the role of the CFO in adopting the right solutions and evolving into a data storyteller with the help of technology, the fear and resistance to change, and the involvement of finance leaders in shaping organizational culture. He also addressed the role of AI agents in handling complex tasks and managing fiscal digitalization.
With over 20 years of experience in finance and software development, Bogdan founded Profluo in 2016, driven by his passion for integrating technology into financial processes. Profluo is transforming the intelligent processing of accounting documents and addressing a major need for efficiency within finance and accounting departments.
You can find the Romanian version of the interview here.
Bogdan, you have a career of over 20 years in finance, going through stages from planning to management in top companies. From this dual perspective, as a finance professional and tech entrepreneur, how do you see the transition from the era in which automation meant only rigid rules in ERP to the current era of generative and adaptive AI?
I believe the transition will be extremely disordered, both as a process and as a final result.
Rigid rules will remain, for two reasons. The first is that many compliance processes are carried out based on them, which are neither optional nor flexible. The second is a pragmatic one: from a computational standpoint, verifying a fixed rule is always cheaper than a heuristic evaluation or a variable reasoning. Therefore, where the nature of the processes requires it, fixed rules will continue to be preferred.
In all other situations – especially where there is ambiguity, a high degree of complexity, or the need for significant cognitive effort – artificial intelligence will be preferred, because it can generate significantly better results than equivalent human effort, at a fraction of the cost.
Managing such a transition is delicate and requires a deep systemic understanding, both of financial cycles within companies and of AI technologies. Here we make a real difference for our clients, because we deeply understand both domains.
There is increasing discussion about the transition from Software-as-a-Service, where the human operates the software, to Services-as-Software, where AI delivers the final result. If we look at the finance department, how does this model redefine the daily activity of a team? Are we still buying tools or are we buying completed processes directly?
Certainly, large clients are no longer looking for AI demos or “toys”, but for concrete results, accompanied by guarantees and high standards of quality specific to professional services.
In this context, we can say that finance departments are no longer buying tools, but complete processes. And if these processes also become relatively autonomous, the role of the finance team moves more and more from execution to supervision, monitoring, adjustment and training of AI models.
These are activities with a high degree of complexity, which require high qualification and, in many cases, bring a more valuable and more interesting professional dimension than traditional operational work.
You said at the CFO Conference Bucharest that “the frontier of artificial intelligence is moving at a fantastic speed” and you mentioned AI agents. For a financial director who still associates AI only with data extraction, can you tell us what an AI agent actually does in a finance department?
Let’s build on the previous idea, that AI agents become truly valuable where data is ambiguous, incomplete, complex or difficult to interpret through fixed rules. In other words, they do not only replace repetitive work, but intervene exactly in those areas where contextualization, correlation and applied judgment are needed.
Transactional activities can be assisted by AI agents in processes such as:
- extracting data from unstructured documents
- normalizing non-uniform data – for example, by correctly identifying items in product catalogs or recognizing incomplete delivery addresses
- accounting and automatic coding – from cost centers, projects and locations, to SAF-T, NC-8 or CPV codes
- correlating documents for making simple decisions – such as matching a receipt with the corresponding invoice, a card voucher with a fiscal receipt, or allocating a payment to multiple sales invoices
- automatic approval of correct transactions
- identifying anomalies – for example, when prices in an invoice do not respect the agreed contract with the supplier and escalating these exceptions to human operators
For reporting and analysis activities, AI agents can bring value through:
- preparing sensitivity analyses and what-if scenarios
- automatically performing variance analyses
- automatically generating reports in specific formats, adapted for certain audiences
In practice, however, reporting and data analysis can be efficiently assisted by AI only if the input data is well structured, normalized and constantly cleaned. That is why it is difficult to imagine a solid agentic flow in the reporting area that is not preceded by a robust agentic flow in the transactional area. In other words, intelligent analysis starts with correct data, and correct data depends on good automation from the beginning.
You also said that “Whoever does not use AI will fall behind – behind their own ego from two months ago.” In a legislative and technological environment that moves so fast, how can a financial leader keep their team relevant without people feeling overwhelmed by the speed of transformation of the solutions they use?
I will give the example of our own company, where we introduced the use of AI agents in the software development area 6 months ago. If the initial reaction was “Do you want to fire us?”, today the conclusion is, I quote: “I don’t think we have any chance in the future without AI.”
For us, the solution was to provide extended access to preferred AI agents to all employees in the software development teams. The result was an explosion of productivity: projects that would normally have required 6-7 months of analysis, planning, development and testing were completed in 1-2 months. And the potential is not exhausted; on the contrary, I believe results can become even better as we integrate AI development agents more deeply into our production flows.
The same principle can be applied by any financial leader: to provide their team with access to AI agents specialized in financial flows and to support them in integrating these new tools into daily activity.
It is, in fact, a gain for everyone. For the company, because it obtains better results, faster and at lower costs. For employees, because they have the chance to increase their contribution without disproportionate effort, to orient their careers towards higher value-added activities and to protect their health from operational transactional stress. And for the CFO, because they respond simultaneously to a double mission: to build a motivated team and to be a role model leader, delivering high-performing, efficient and sustainable financial services to the organization.
One effect of these rapid changes is the increase in fear among employees, that concern that they will be replaced by AI. From the experience of Profluo implementations, what is the optimal strategy through which a CFO can communicate to their team that AI agents are not replacements, but partners in daily activity? And further, where does human activity move when people are freed from repetitive, time-consuming tasks?
In mature teams, operators want routine work to disappear, and their role to focus on supervising process quality and analyzing results – that is, exactly those truly senior activities. This aspiration is fully aligned with the CFO’s objective, which seeks a higher level of quality, accuracy and scalability of financial services delivered to the organization. For this reason, adoption can be natural and even enthusiastic.
In less mature teams, however, artificial intelligence is often perceived as an existential threat, to which operators react through direct opposition or sometimes through more subtle forms of resistance.
A good CFO will manage this resistance as they approach any successful transformation project: through careful communication about what the future of the team looks like, through direct involvement and through the power of personal example.
In this context, a mature Finance AI technology provider can make the difference, quickly demonstrating good results for both “sides” and facilitating the transformation of resistance into cooperation.
Taking the discussion further, what skills should a member of a finance team develop today to remain indispensable in the future? And what about CFOs?
I believe the need for technical expertise in the financial field will remain, but it will evolve to an increasingly higher level of sophistication. AI agents can take over high-volume, high-frequency transactional tasks, but, at least in finance departments, they will continue to need human supervision, validation and filtering.
The reason is simple: fiduciary and professional responsibility will remain, essentially, the responsibility of people. You cannot hold an AI agent accountable for a wrong decision, just as you cannot justify in front of a board of directors a wrong decision by simply saying it was your AI agent’s recommendation.
As AI agents’ capabilities advance, the same must happen with the competence level of those who supervise them. In other words, AI evolution does not reduce the need for human expertise, but pushes it toward a more advanced, more critical and more responsible form.
In a company where processes are carried out by AI agents that do not take breaks and do not get tired, how does human team culture change? How can financial leaders maintain team spirit and a sense of belonging when some “colleagues” are actually digital entities?
I am very optimistic. I have seen many teams where the permanent stress generated by transactional volume, the pressure of correcting errors and the urgency of deadlines have led to an unbalanced work climate. I am convinced that reducing this type of pressure, through the use of AI agents in transactional activities, can have a profoundly beneficial effect on how teams work and collaborate.
The culture of the future will place greater emphasis on the quality of organizational objectives, business results and human relationships.
At the same event you mentioned a challenging aspect: the cost and efficiency of onboarding a junior versus an AI agent. If “apprenticeship work”, which used to involve manual processing and data entry, is taken over by AI, how do we train the next generation of experts? Where should a young finance graduate start today?
I have a mixed feeling, between concern and hope. There is already a new term, “cognitive rust belt”, which describes the erosion of people’s capacity for analysis, critical thinking and judgment as more and more tasks are delegated to AI agents.
The risk is real. But it can be countered precisely by equipping people with their own AI agents and continuously rebuilding human expertise in the domains they choose in their careers. The combination of human and agents is unbeatable, which is why it must be cultivated with a strong emphasis on human development.
I believe this is one of the few ways in which human value can not only be protected, but truly increased in the society of the future.
That is why any young finance graduate should choose a preferred financial field in which to specialize, and then train their own AI agent so that it helps them become exceptionally good in that direction. The combination of the two can make them relevant, competitive and employable anywhere and anytime.
Many companies believe their internal processes are too specific or too complex to be standardized by external technology. From your experience, where is the line between adapting technology to company habits and changing the organizational mindset to adopt innovation?
I believe that, in the case of classical technologies, customization at scale and managing complexity were real barriers. In many situations, companies were right to complain that, in order to adopt standardized technology, they had to “mutilate” their own internal processes.
Cognitive technologies fundamentally change this reality, because they solve adaptability natively. An AI agent can replicate the flexibility of a human operator: it can receive specific instructions on how to execute a process, it can semantically understand documentation and can autonomously derive its own workflow based on set objectives.
From this perspective, the problem of adapting technology to company processes will significantly decrease. However, one major challenge will remain: organizational resistance to change.
In discussions about AI, the question often arises: “What is the concrete business impact?” From Profluo implementations, what measurable results does a CFO see (in terms of cost, time or accuracy) and how quickly?
Certainly, we need to measure the business impact of any initiative, even when it seems logical and easy to adopt.
For example, in our software development activity, a concrete result of implementing AI agents was the increase in code delivery capacity and the acceleration of project execution pace. Our latest analysis shows that, at least in the architecture area, 87% of the code is written by agents.
At client level, there are several highly relevant performance metrics:
- the average speed of taking, extracting data, validating, fully accounting and posting a financial document into ERP is under 5 seconds
- the accuracy of the entire flow exceeds 98%
- the implementation and configuration speed of the Profluo platform is measured in hours, sometimes even minutes
Fiscal digitalization (e-Invoicing, e-Reporting) is radically changing how finance departments operate. How do you see AI managing this complexity and turning compliance from an obligation into an operational advantage?
This is an extremely hot topic across Europe and beyond.
If a traditional invoice has around 40 fields, a European standard e-invoice reaches approximately 160 fields, while the Polish e-invoice exceeds 400 fields. Additionally, the upcoming implementation of e-Invoicing in France includes specific e-Reporting requirements, including intermediate transmission statuses and even payment statuses.
This clearly shows that, despite standardization efforts, each jurisdiction continues to add its own requirements.
The first reaction of many companies is urgency: to solve the immediate problem. In other words, to adopt a simple “XML postman” that pushes invoices into ERP. This is exactly what happened in Romania and we see the same pattern elsewhere.
After this urgency passes, a shift in perspective emerges. Companies begin to understand that e-Invoicing and e-Reporting are not just compliance obligations, but the financial infrastructure of the future. Once you have large volumes of structured data, you can build complete agentic workflows on top of them.
What are the most frequent mistakes you have seen in financial automation projects?
The main mistake is starting to buy AI agent platforms just because they are trendy. Without authentic motivation and a rigorous, measured approach, implementation is likely to fail.
The second major mistake is being surprised by resistance to change. It is natural and must be addressed, but not passively accepted.
Almost no organization transforms spontaneously from within. If leadership does not drive change, it will come from external shocks, which are usually far more severe than proactive transformation led by a financial leader.
