Mass-scale, routine-based work, which some companies have outsourced to BPO (Business Process Outsourcing), is now facing a pivotal question: what next? BPO service providers profited precisely from this: from volume, from the fact that it was performed by moderately skilled employees, and from taking responsibility for service quality and execution. But what should be done now, when simple, repetitive processes can be handled by AI-based solutions?
What does BPO really sell and why is AI targeting it?
Companies entrust service providers with responsibility for quality, team management, team rotations, and often also the technology necessary to perform specific tasks. This allows them to focus on achieving business goals. By outsourcing processes, you free up time, of course, for a certain price. This business model worked perfectly until more advanced RPA solutions were implemented. I observe that clients are starting to question why they should pay for services when they can take over the processes themselves. The motivation to change the model is further reinforced by the availability of AI-based tools, as we have already written about regarding process automation in companies. BPO services can be performed using them. So why pay for outsourcing?
Which processes previously outsourced to BPO can AI now handle?
AI most quickly takes over what BPO handled on a massive scale and according to a repetitive pattern. Gartner predicts that by 2029, AI will autonomously resolve 80% of common customer service issues, resulting in a 30% reduction in operational costs. This is just one example of BPO services that will be displaced by AI.
Accounting Services Outsourcing
Electronic invoice workflow, structured documents, widespread OCR technology, and now the potential application of AI in more advanced financial and accounting processes create opportunities to reduce BPO costs. Giving up a provider in favor of investing in modern tools offers dual benefits. Firstly, financial, and secondly, related to possessing modern technology that can be independently developed or easily adapted to one's own needs.
HR and Payroll Outsourcing
Collecting and validating input data for payroll, handling typical employee requests for leave, certificates, or salary components, and preliminary data consistency checks before calculation are ideal examples of processes where AI can excel. The employee's role involves final approval, which in the case of payroll was typically already on the side of the company commissioning the service.
Back Office Process Outsourcing
Transcribing from documents to systems, completing records, migrations, administration and handling of documentation – this was a daily routine in BPO. This scope of duties can also be replaced by artificial intelligence-based tools.
What AI won't take over from a provider and when is it worth sticking with BPO?
Outsourcing processes to an external company makes sense where it involves taking responsibility, not just "handing over" the performance of tasks. The BPO provider assumes responsibility for regulatory compliance, deadlines specified in the SLA, and financial consequences if the process fails. AI will not make a final decision, send a declaration without a signature, or independently respond to a letter from the Social Insurance Institution (ZUS) or the tax office. These activities still require human involvement.
The second case is when a process is not repetitive and schematic, requires an individual approach and judgment, or human interaction. Conducting conversations or negotiations are still tasks performed by people.
The third case concerns activities that, according to the AI Act, cannot be fully performed by AI-based tools. These are primarily decisions concerning people, which the AI Act classifies as high-risk decisions and requires human oversight. These include:
- HR decisions regarding recruitment
- employee evaluations or promotions
- financial decisions (creditworthiness assessment and access to financial services)
In these areas, an AI system can prepare an analysis and recommendation, but the final decision must be made and approved by a human.
It's also worth sticking with outsourcing when the volume is low and the investment in AI won't pay off, the process is rare or highly variable, regulations require supplier guarantees, or there's a lack of data and a team to maintain the model.
In such situations, outsourcing still wins.
Own AI or continued process outsourcing: how to calculate what pays off?
We calculate the profitability of a project just as we did when deciding on BPO. That is, how much the company will save compared to the investments it has to incur for the purchase, maintenance, and operation of an AI-based solution. Simple calculation, simple ROI from AI implementation. Sometimes, however, it is not financial benefits and calculations that determine the optimal solution, but the required commitment and knowledge. Added to this are the risks associated with maintaining models and tools, data storage, and cybersecurity. Sometimes, issues related to security and required commitment are more important than purely financial considerations.
How to transfer service outsourcing from a provider to your own AI without downtime?
Transferring a process from an external provider to your own AI without the risk of downtime requires a project-based approach and planning the entire transition in an overlapping manner. Similarly to transferring processes to a provider, where a so-called "transition" period was necessary, a time during which two teams worked simultaneously. In the case of implementing AI tools, particularly in processes related to HR, payroll, and finance, where legally defined deadlines apply, ensuring continuity is critical. Artificial intelligence and RPA tools must be tested and verified to eliminate operational risks.
Summary
In my opinion, artificial intelligence has significantly changed the BPO market, taking over mass-scale, repetitive, and procedure-based processes in accounting, HR and payroll, and back office from external providers. For businesses, this creates an opportunity to reduce costs and build their own flexible technological competencies.
However, transitioning from outsourcing to internal AI is not always justified. BPO still wins where legal expertise and regulatory compliance, as well as responsibility and risk, are crucial.
If you're wondering which processes in your company are suitable for AI takeover today, and which should remain with a provider, this decision rarely boils down to just the tool. More often, it's about whether the organization and team are ready to take on the responsibility that the BPO provider previously assumed. Check where you stand on this scale before you start by purchasing technology.
