The Spanish Customer Service Act (Law 10/2025) has brought to light a reality that many companies were already aware of: effectively managing customer service isn’t just about having the right channels or an internal procedure in place. You also need to be able to understand what’s actually happening through those channels and, when the time comes, prove it.
The law establishes minimum quality standards for certain customer service offerings and applies to both companies providing basic services of general interest and large companies falling within its scope.
Among other issues, it regulates response times, access to personalized service, complaints, accessibility, and languages. These obligations have a direct consequence for organizations: compliance requires data and evidence.
And that’s where artificial intelligence begins to play an interesting role. Not because an AI tool alone can guarantee that a company complies with the law, but because it can help monitor in a much more systematic way what happens in thousands of calls, complaints, and interactions that until now were difficult to oversee.
The SAC Act turns compliance into a data problem
Many of the obligations established by the SAC Law can be expressed through objective indicators. For example, the law stipulates that 95% of incoming calls must be answered, on average, in less than three minutes. Furthermore, when a customer requests personalized assistance, 95% of these requests must also be addressed, on average, within three minutes of being made.
The law also prohibits the use of automated answering systems, chatbots, or other similar systems as the sole means of customer service. Customers must be able to request assistance from a live agent through the options available on the main menu and during the interaction.
Added to this are other requirements, such as those related to complaints and incidents. As a general rule, inquiries, complaints, claims, and incidents must be resolved within a maximum of 15 business days, while those related to billing or improper charges have a maximum resolution period of five days. There are also specific obligations for certain incidents related to essential services.
This raises a fundamental question: How can a company demonstrate that it is consistently meeting these requirements? It is not enough to have a written procedure; it is necessary to know what is actually happening during calls, with complaints, in customer service systems, and in internal processes—and that means working with data.
From call sample monitoring to comprehensive oversight
For years, a significant portion of quality control in contact centers has relied on listening to call samples. This is a logical approach when analysis depends exclusively on people. If an organization handles hundreds of thousands of conversations per month, it is physically impossible to listen to them all. A sample is selected, evaluated, and used as a benchmark for what is happening.
The problem arises when that sample is also used to monitor compliance risks. It can provide a good snapshot of certain situations, but there’s always the possibility that problematic cases fall outside the selected calls.
Speech analytics and AI-powered natural language processing technologies allow us to approach this task differently. A conversation can be automatically transcribed and analyzed to identify specific elements: what language was used, what the reason for the contact was, whether the customer asked to speak with a person, whether a specific mandatory disclosure was made, or whether a complaint arose during the call.
This does not mean eliminating human review. It means using it where it truly adds value. Instead of devoting resources to manually searching for potential deviations, technology can help identify interactions that require attention and then facilitate their review by the teams.
It’s a fairly significant shift: moving from looking for problems within a sample to using the data to identify where it’s worth looking.
Translating the SAC Act into measurable indicators and evidence
One of the most practical uses of AI is precisely to translate certain regulatory requirements into indicators that can be continuously monitored.
Not every provision of a law can be turned into an automated rule, nor would it be reasonable to try. Some issues require legal interpretation, and in others, human intervention will remain essential.
But there are many operational aspects that can be measured. For example, average handling times, requests for personalized assistance, transfers made, pending complaints, resolution times, requested language, or the availability of evidence associated with a case.
From there, you can build a control panel that allows you to identify which requirements are being met and, above all, which ones are beginning to deviate. The difference from a traditional dashboard lies in traceability. An indicator showing 93% compliance is useful. But it immediately raises another question: What happened with the remaining 7%? If the system allows you to drill down from that percentage to the interactions that explain it, the call, the transcript, the CRM record, or the corresponding timestamp, the indicator ceases to be merely a metric and becomes a tool for investigation and auditing.
That level of detail can be especially valuable when analyzing a deviation, responding to a complaint, or documenting how a specific interaction was handled.
Using AI to prevent, not just to audit
So far, we’ve been talking mainly about monitoring: analyzing what happened to verify whether the service is functioning as it should. But AI can also intervene earlier, and a good example is demand spikes. If a company discovers the next day that wait times skyrocketed for two hours, the analysis will have served to detect the problem, but not to prevent it.
Here, conversational agents can play another role. An AI agent can initially assist the customer, identify the reason for their call, gather the necessary information, and determine which team to direct them to. If human intervention is needed, the conversation can be transferred while providing the agent with the previously gathered context.
It can also be useful for managing call surges, identifying priority situations, or locating an agent who can assist in a specific language. That said, automation should not be viewed as a choice between AI and human agents.
The SAC Law permits the use of automated answering systems, conversational bots, and similar systems, but requires that they allow customers to request personalized assistance from the main menu at any point during the interaction.
Therefore, the most useful approach may not be to view AI as a substitute for agents, but rather as a technology capable of better organizing demand and facilitating the transition between automation and human assistance. The goal is for technology to facilitate access to the service, not to create new barriers.
From reactive compliance to proactive Compliance
The implementation of the SAC Act is forcing many companies to review their customer service processes. But simply checking off a list of requirements likely only scratches the surface of the problem.
There’s an even more interesting opportunity. If an organization is able to connect its conversations, its operating systems, and its compliance metrics, it can begin to detect problems before they become systemic.
It can identify that certain hours are causing excessive wait times; that a specific type of complaint is taking too long to resolve; that some transfers are leading to a poor customer experience; or that certain procedures aren’t being followed as intended.
That’s where AI adds the most value: not simply by generating a report at the end of the month, but by helping to identify, explain, and correct deviations.
Of course, technology is no substitute for governance, legal judgment, or oversight by the responsible teams, nor should everything that can be automated be automated. But it does change the scale at which monitoring can take place.
For a long time, companies have had to rely on samples, periodic audits, and data scattered across different systems. The combination of conversational analytics, AI, and data integration allows us to move toward a different model: compliance that is more continuous, traceable, and connected to day-to-day operations.
And perhaps that is one of the most significant consequences of the SAC Act. It’s no longer just about having the right processes in place. It’s about being able to verify that they work, detect when they stop working, and have the evidence to prove it.
Learn more about how using intelligent agents can help you optimize your service and comply with regulations by clicking here.