AI agents are a strategic competitive factor. Many companies that are already using generative artificial intelligence (AI) will introduce AI agents in 2025. But how to build a reliable and secure system?

The benefits of multi-agent systems are obvious: they can autonomously and effectively perform tasks within an organisation. In marketing, for example, they do research, provide insights and assist with content creation, design and customer service. Not only is this efficient, but it also reduces the workload on employees. It is therefore likely that AI agent systems will become commonplace in all sectors, from retail, insurance, finance and healthcare to logistics, tourism and industry. And this is not just a future dream: AI agents are already helping with customer service at Hey Lou Hotels, planning air cargo handling at Frankfurt Airport, and making OpenTable reservations.

As the agents act completely independently, they must be able to act and communicate responsibly. And they must not violate data protection and compliance guidelines. It is therefore essential to build multi-agents on the basis of corporate values. 

Blockchain as the basis for security, transparency and reliability

Many companies already have ethical guidelines for their use of Gen AI. These include aspects such as security (privacy and compliance), integrity (transparency and accountability) and reliability (control and robust systems). They know that this new technology is not just about profitability, but also about trust.

But how can companies be sure that their AI agents are performing tasks for customers in an ethically correct manner, such as making flight, hotel or restaurant bookings, recommending investments in financial products or reporting claims? Serviceplan Group has developed a reliable solution for this: the blockchain protocol Masumi. Each AI agent in the protocol is given an identity through a decentralised identifier (DID), and logs immutable cryptographic hashes of its results and transactions on a blockchain. This allows Masumi to uniquely determine the identities of AI agents, record their transactions, and hold them accountable for their decisions and actions. By identifying each individual agent, managers can track their activities and information flows. Blockchain technology provides the desired transparency and control - and ultimately the reliability of the system.

Privacy and security in trusted hands

But the technology can do even more: because blockchain is based on a decentralised system, it is easier to defend against hacking and manipulation. This is particularly important in multi-agent systems, where the integrity of the information exchanged between agents is at times critical. After all, they rely on internal tools and data sources to perform their tasks.

Masumi has introduced additional standards, such as verifiable proof of authorisation, which ensures greater security: before an agent can access CRM data, for example, they receive a precise authorisation that can be checked, changed or revoked by the responsible person. Techniques such as Zero-Knowledge (ZK) proofing ensure that no sensitive or personal data is shared. For example, a LinkedIn agent can post something to a personal profile without revealing the underlying authorisation details or the identity of the owner.

The choice of the blockchain is crucial

The Masumi blockchain protocol ensures that ethical guidelines and legal regulations are adhered to. These include regulations such as GDPR (General Data Protection Regulation), MiCA (Markets in Crypto-Assets Regulation), KYC (Know Your Customer) and AML (Anti-Money Laundering).

The choice of blockchain is crucial in order to take ethical guidelines into account. The Cardano blockchain, on which Masumi is based, features the Proof-of-Stake (PoS) consensus process, which is one of the most energy-efficient in the blockchain world and eliminates the need for energy-intensive mining. In addition to sustainability, Cardano's multi-layer architecture is also an important advantage. On the one hand, it enables massive parallelization, allowing the blockchain to handle large volumes of simultaneous interactions, and on the other, it guarantees a high level of security.

Masumi creates an infrastructure that makes agent interactions verifiable, secure, private and legally compliant. In this way, the Serviceplan Group ensures that multi-agent systems can be integrated into companies in an ethically correct, responsible and trustworthy manner.

Author: Sebastian Küpers, Chief Transformation Officer, Plan.Net Group