Why Agentic AI Is Non-Deterministic (and Why That Matters)

Traditional code is deterministic: the same input always produces the exact same output. Agentic AI, powered by large language models (LLMs), is not. LLMs generate responses probabilistically – they sample from a distribution of possible tokens based on patterns learned during training. Even with identical prompts, outputs can vary due to:

• Temperature and sampling parameters
• Multi-step reasoning loops (planning → tool selection → execution → re-evaluation)
• Hallucinations or subtle prompt sensitivities
• External tool results that feed back into the next reasoning step

When an agent is given the ability to act (send an email, post in Slack, query a database, or trigger a business process), this non-determinism creates real risk. A seemingly harmless prompt can lead to prompt injection, tool misuse, unintended data exposure, or actions outside the intended scope. The system is no longer a predictable calculator – it becomes an autonomous decision-maker operating in a probabilistic space.

That is exactly why Agentic AI must be controlled tightly. Without deliberate, layered safeguards, the blast radius of a single unexpected output can be 

Two Secure Paths to Agentic AI

We guide clients based on their needs and technical maturity:

1. Vendor-Managed Walled-Garden Platforms (ideal for non-technical teams)
For business users who need AI chat, workflows, or integrations (Slack, Gmail, internal docs) without building infrastructure, we recommend fully sanctioned, vendor-managed platforms that meet our strict compliance bar:

Key requirements we enforce:

• ISO 27001, SOC 2 Type II
• Full GDPR alignment
• EU hosting
• No training on customer data
• Transparent sub-processors and DPA
• Built-in (or easily configurable) human-in-the-loop (HITL), confirmation steps, and basic guardrails

Currently under final evaluation:

• Langdock (Berlin/Azure EU):  strong for workflows and agents with Guardrails node, per-action confirmation (HITL), and admin controls.
• Lurus (Hannover):  affordable multi-agent automation for chat-initiated tasks.

The core rule is simple: stay inside the platform. The vendor owns the security model –  but only within their boundary. We help clients configure these platforms (admin hardening, workflow design, HITL policies) so they remain safe and auditable.

2. Custom Bounded Agents (for teams needing maximum control)
When pre-built integrations aren’t enough, we develop fully controlled agents using a robust defense-in-depth stack:

• Orchestration with LangGraph and Inngest
• Guardrails AI for robust validation
• Lasso for traffic inspection
• ToolHive for per-tool isolation

These agents feature narrow tool scopes, comprehensive auditing, and human oversight gates — delivering precision and security.

Datarella’s Agentic AI Services

• Strategy & Architecture: Risk assessment and governance frameworks
• Platform Selection & Hardening: Compliant vendor evaluation and secure configuration
• Custom Agent Development: Production-grade bounded agents
• Tool & MCP Security:  Secure integration of external capabilities
• Enablement & Assurance: Team training, guidelines, and ongoing audits

Why Partner with Datarella?

Agentic AI unlocks major productivity gains, but its non-deterministic nature demands rigorous control. We combine deep expertise with practical, EU-compliant solutions so you can deploy agents confidently –  whether through trusted vendor platforms or tightly bound custom systems.

Ready to implement secure Agentic AI? Contact Datarella today to discuss your use case.

The post Agentic AI at Datarella: Secure Intelligence for Enterprise Workflows appeared first on DATARELLA.

In April 2026, two timely pieces cut through the AI hype cycle. Ethan Mollick, writing in The Economist, warned that “the IT department [is] where AI goes to die.” His core argument: AI is a profoundly odd, risky, and powerful technology, a next-word predictor that unexpectedly writes code, gives strategic advice, or shows empathy, and companies are killing its potential by trying to “de-weird” it. Traditional IT processes, risk-averse governance, standardized KPIs, and legacy systems force AI into the mold of conventional enterprise software, stifling experimentation and emergent value. 

At the same time, the Financial Times published “Investors are betting on AI chaos. History suggests otherwise.” Author Richard Waters noted that markets are pricing in revolutionary disruption – new winners, old losers – but past technology revolutions (PCs, internet, cloud) show savvy incumbents often muddle through, adapt, and even thrive. The real story is rarely the total chaos investors crave. 

Together, these perspectives paint a clear picture of the most likely future of AI in business: not a dystopian job apocalypse or unicorn-disrupting chaos, but a pragmatic, evolutionary integration. AI will augment human work, reshape workflows, and deliver real value, but only for organizations that treat it as the strange, probabilistic tool it is while building the right foundations. Incumbents with strong data capabilities will have the edge.

Why AI “Dies” in Traditional IT, and Why That’s the Wrong Approach

Mollick’s essay resonates because it diagnoses a widespread problem we see daily in enterprise deployments. AI isn’t deterministic software with predictable outputs. It’s generative, context-dependent, and often surprising. When companies hand it to IT teams focused on security, compliance, uptime, and cost control, the natural response is to:

• Wrap it in rigid approval processes
• Demand ROI projections before pilots
• Force it into existing tech stacks without workflow redesign
• Prioritize “safe” use cases over creative experimentation

The result? Pilots that never scale. According to recent analyses (including Deloitte’s 2026 State of AI in the Enterprise), while worker access to AI tools has surged, the number of organizations moving projects into full production remains modest. Barriers like data quality, skills gaps, infrastructure readiness, and risk management continue to stall progress.

HBR has similarly observed that many companies report “widespread AI usage but disappointing returns,” with adoption stalling at the integration stage. The problem isn’t execution – it’s treating AI like a new CRM module instead of a fundamentally new way of working. 

History Shows Incumbents Can Win If They Adapt Smartly

The FT piece offers reassurance: AI won’t necessarily destroy every incumbent. Past waves of technology (from electricity to the internet) initially sparked predictions of massive disruption, yet established players who invested in complementary capabilities—new skills, processes, and organizational structures—came out stronger.

In 2026, the winners won’t be the pure-play AI startups alone. They will be enterprises that:

• Combine their domain expertise and proprietary data with AI’s capabilities
• Redesign workflows around human-AI collaboration (what some call “co-intelligence”)
• Scale from pilots to enterprise-wide agentic systems under proper guardrails

PwC’s 2026 AI Business Predictions and similar reports emphasize a “disciplined march to value”: top-down enterprise strategies, measurable business outcomes, and governance that doesn’t kill experimentation. 

The Most Likely Future: Pragmatic, Data-Driven, and Agentic

By late 2026 and into 2027, we expect the following trajectory:

1. From pilots to production at scale — Organizations doubling the share of AI projects in production, driven by agentic AI (autonomous agents that execute multi-step workflows).
2. J-curve productivity — Initial flat or negative returns as companies rewire processes, followed by steep gains once complementary innovations (new roles, data pipelines, decision protocols) are in place.
3. Governance catching up — Mature frameworks for agentic AI, data quality, and responsible use becoming table stakes. Shadow AI will decline as secure, enterprise-grade platforms mature.
4. Incumbents leveraging data moats — Companies with clean, governed data and domain expertise will outperform pure AI-native players in regulated or complex industries.

This future is neither utopian revolution nor failure – it’s an evolutionary transformation, provided organizations avoid the “IT department trap.”

Optimal Use of AI in Businesses: Five Practical Principles

Drawing from Mollick, historical lessons, and 2026 enterprise reports (Deloitte, McKinsey, PwC), here’s how forward-thinking companies are winning:

1. Embrace the weirdness—experiment deliberately
   Give teams space to discover unexpected uses. Mollick advocates leadership that encourages crowdsourced experimentation and “labs” to scale promising ideas. Treat AI like a creative collaborator, not just an automation tool.
2. Build on rock-solid data foundations
   Data quality and governance remain the #1 barrier cited across reports. Without trustworthy data pipelines, AI outputs are unreliable. This is where specialized partners excel – unifying siloed data, implementing real-time pipelines, and ensuring privacy/compliance.
3. Redesign workflows and roles around human-AI co-intelligence
   Don’t automate jobs – augment them. Successful organizations are re-architecting processes so humans focus on judgment, creativity, and relationships while AI handles analysis, drafting, and routine execution.
4. Deploy secure, governed agentic AI
   Autonomous agents are the next frontier, but they require bounded orchestration, threat modeling, and compliance-by-design. Enterprises need platforms that support multi-agent systems without introducing new risks.
5. Measure what matters – and iterate
   Move beyond vanity metrics. Track business outcomes (revenue impact, cost savings, customer satisfaction) and accept that ROI may follow a J-curve.

How Datarella Helps Businesses Navigate This Future

At Datarella, we’ve spent years helping organizations move beyond AI hype and pilot purgatory. Our expertise in AI agent development and security, full-stack application modernization, Web3-enabled decentralized solutions, and privacy-preserving data architectures directly addresses the challenges outlined above.

Whether you need:

• Secure, production-ready autonomous agents
• Data platforms that make AI reliable and compliant
• Integration of AI into legacy systems without the usual friction
• Or decentralized approaches that enhance trust and data integrity

We combine deep technical capability with practical business understanding to help you safely and scalably embrace AI’s weirdness.

The future of AI in business isn’t about replacing your IT department or betting everything on chaos. It’s about evolving how your organization learns, decides, and creates value—by treating AI as the strange, powerful tool it is, while building the data, governance, and cultural foundations it demands.

Ready to move from pilots to production without letting AI “die in IT”? Let’s talk. Contact Datarella to explore how we can help you capture the real, pragmatic upside of AI in 2026 and beyond.

The post The Most Likely Future of AI: Embracing Its “Weirdness” Without Chaos appeared first on DATARELLA.

Building the Foundation for Trusted Machine Communication

Working alongside industry leaders such as Bosch, Airbus, Continental, and leading Web3 projects, we contributed to building the technical and conceptual foundation for trusted machine-to-machine communication in the mobility sector. This included the secure exchange of credentials, decentralized data marketplaces, and AI-powered autonomous service interactions—all compliant with European data and privacy standards.

Demonstrating Real-World Impact: MOBIX Park & Charge

A standout achievement was the development and public demonstration of MOBIX Park & Charge, a fully operational system enabling electric vehicles (EV) to autonomously find parking spots, access charging stations, and handle payments. First showcased at IAA Mobility 2023, the system integrated SSI, blockchain-based payments, and AI agents in a live, real-world environment.

Scaling Toward Smart Cities

Beyond the demo, MOBIX has evolved into a scalable smart-city solution. By turning private EV chargers and parking spots into publicly accessible assets, we’re addressing key challenges in urban congestion and infrastructure scalability, while opening up new economic opportunities for individuals and municipalities.

Where Web3 Meets AI

The project also served as a powerful example of the convergence between AI and Web3. By combining intelligent agents with decentralized infrastructure, we demonstrated how machines can not only interact but also negotiate, transact, and self-optimize—laying the groundwork for more ethical and transparent digital ecosystems.

A Foundation for the Future

In sum, moveID wasn’t just about mobility. It showcased how decentralized identity, autonomous agents, and AI can reshape how devices, services, and users interact—not just in cities, but across industries. As the project concludes, its outcomes provide a strong foundation for future applications in smart infrastructure, data sovereignty, and the broader digital economy.

The post moveID: Wrapping Up Three Years of Mobility Innovation appeared first on DATARELLA.

As noted by the venture capital firm Andreessen Horowitz, AI and Web3 are converging technologies that leverage decentralized networks, self-sovereign identities (SSI), autonomous agents, and machine learning. These innovations enable new business models and streamline business processes, particularly in the realm of decentralized mobility solutions. A crucial success factor lies in integrating these technologies with existing business processes. Collaboration between innovative AI and Web3 startups and established industry leaders is essential for driving impact and fostering innovation.

At IAA Mobility, AI and Web3 startups collaborate with industry players and universities in the moveID consortium. They are showcasing a state-of-the-art smart city solution called MOBIX Park & Charge. To meet the project’s goals, positioning innovative cutting-edge solutions is vital. These solutions leverage data to redefine the technological landscape, heralding a data-driven revolution. The synergistic trio of Fetch.ai, a long-time partner of Datarella, along with peaq and Ocean Protocol as moveID partners, is making significant strides forward. This initiative is further supported by solution providers and integrators like Datarella, DeltaDAO, 51Nodes, and university teams from htw saar and Zeppelin University.

Fetch.ai: Bridging AI and Blockchain

Datarella partner Fetch.ai seamlessly combines artificial intelligence (AI) with blockchain technology, unlocking new possibilities across various sectors. These possibilities include autonomous machine economies, smart cities, and efficient resource management. The Fetch.ai platform provides a decentralized infrastructure for Microagents capable of executing tasks autonomously.

In the context of MOBIX Park & Charge, Microagents act as bridges between different software components on edge devices. This setup enables autonomous control over various systems, such as traffic lights.

Microagents in Action

Fetch.ai Microagents serve as multifunctional intermediaries. They communicate and initiate processes in external systems, handling complex tasks like:

• Managing crypto wallets
• Ensuring secure payment transfers
• Verifying events
• Controlling access to hardware such as gates, traffic lights, and chargers

In essence, Fetch.ai Microagents automate tasks on behalf of human clients, streamlining processes and enhancing efficiency. These functionalities contribute significantly to the development of decentralized mobility solutions.

peaq: Empowering the Economy of Things

As automation surges, the peaq network emerges as a key player in sharing its benefits with everyone. It fosters the Economy of Things by establishing a crucial layer-one blockchain infrastructure for decentralized mobility applications.

Peaq’s technology stack and economic incentives enable the development of applications for machines in this emerging economy. For instance, it powers electric vehicle (EV) charging applications. The peaq token fuels the entire ecosystem, facilitating digital identities for machines and enabling seamless payments.

In MOBIX Park & Charge, peaq allows communication with charging points, using peaq tokens to initiate and pay for EV charging processes.

Ocean Protocol: A New Data Economy

Ocean Protocol aims to unleash its potential for individuals and organizations alike. It empowers secure data sharing (Compute-to-Data), selling, and monetization while ensuring control and privacy. Users retain the power to determine who accesses their data and how it is used.

This establishes a novel paradigm in the data economy, especially for AI models in the mobility sector. In MOBIX Park & Charge, Ocean Protocol integrates its decentralized data marketplace technology. This facilitates sovereign and privacy-preserving data exchange within mobility applications.

Key Components of Ocean Protocol

Key components of Ocean’s data pricing mechanisms include Compute-to-Data and a Gaia-X-compliant Self-Sovereign Identity (SSI) approach. Additionally, Ocean solution provider DeltaDAO collaborates to implement the Ocean Tech Stack. In future versions of MOBIX Park & Charge, the goal is to simplify data sales for IoT devices. These devices will automatically sell data directly on the Ocean Marketplace or through Compute to Data. Fetch.ai agents possess built-in intelligence, allowing IoT devices in cars to decide when and how to post data to the Ocean Marketplace based on traffic situations.

Conclusion: Paving the Way for a Data-Driven Future

The versatile infrastructure of peaq, the automated decision-making capabilities of Fetch.ai’s Microagents, and Ocean Protocol’s sovereign data exchange mechanism synergistically converge to facilitate MOBIX Park & Charge. As we stride toward a data-driven future, these technologies pave the way for decentralized mobility solutions that could profoundly reshape mobility.

The post The Convergence of AI and Web3: MOBIX Park & Charge at IAA Mobility 2023 appeared first on DATARELLA.

Introduction to Ocean Protocol’s decentralized data marketplace

The Ocean Protocol is a decentralized data exchange protocol that aims to unlock the value of data and enable secure and privacy-preserving on-chain data sharing and monetization ⛓️. It provides a framework for building data-centric applications that facilitate the exchange of data assets while ensuring data privacy, security, and compliance.

The Ocean Protocol marketplace is the heart of the entire Ocean Protocol ecosystem. These data sets can be bought and sold along with access to computation on those data. What Uniswap did for Finance, Ocean Protocol is doing for data. Ocean goes beyond just merely trading data in the clear though and makes it possible to order computation on data you never see and get actionable results while maintaining data privacy and secrecy. That in many ways is the holy grail of data privacy and security!

The Ocean Protocol approach to data markets is already gaining substantial traction in industries through the MoveID consortium and GAIA-X where Datarella is a partner. Specifically, there are already industry focused GAIA-X compliant Marketplaces utilizing Ocean protocol already released in the wild (see Pontus-X)

Fetch.ai Agents + Ocean Protocol: A Perfect Match

Switching gears, you may have also heard of Fetch.ai. It provides a next-generation protocol enabling a digital world where Autonomous Economic Agents and Microagents can perform proactive economic activity. AEAs have some unique attributes. Fetch Agents (AEAs) can find resources on their own, they have private keys so they can transact autonomously, and, more importantly, they can decide to execute business logic autonomously based on their programmed skills and behaviors. AEAs are perfect for delegating on-chain tasks that you need to make sure actually happen technically but don’t want to have to handle manually. A great example of this is the stop loss agents that Fetch.ai has released in order to allow people to provision liquidity on DeFi protocols autonomously. Another real-life use case is the kind of behavior that we demonstrated with MOBIX at the International Automobile Show (IAA) in 2021.

In that context, we demonstrated how an Autonomous Economic Agent, running on an edge device and accessible through a specialized version of the MOBIX app, could enable privacy-preserving ad workflows. That sounds super complicated, but its utility is really simple.

Smart Recommendations without Data-Lake-Honeypots

Consumers have come to expect really smart recommendations in their apps based on their current context and previous actions. We’ve come to take this for granted everywhere. It extends from the autocomplete suggestions in your search field to Netflix recommendations to mobility recommendations embedded in scooter rental apps. There’s a shadow side of this intelligence though. Simply put, in order to offer that kind of intelligence someone is invariably building a vast data lake full of your user data to power those convenient algorithms. We all know that this kind of data accumulation has massive privacy ramifications, not to mention the security risks associated with those giant data-lake-honeypots. By connecting Fetch.ai Autonomous Economic Agents with Ocean Compute to Data workflows and the Ocean Marketplace we are laying the cornerstone for enabling that kind context aware convenience without sacrificing any user privacy.

Moreover, outside of the consumer sector in the world of IoT devices, this code makes it really convenient for any number of machines to automatically sell their data either directly on the Ocean Marketplace or via Compute to Data! AEAs have a certain amount of built-in intelligence. This enables IoT devices from cars to weather stations to decide, based on their situation, when and how to post the data to the Ocean Marketplace for sale. Additionally, it allows consuming devices to autonomously search out and purchase the data that they need. Together, this approach provides the basis for autonomous adaptive machine learning without having to own your own ample data sets.

Privacy-Preserving Prediction Engines

TLDR: By enabling Fetch Agents to post data to Ocean Marketplace Compute to Data recommendation engines can be driven without violating privacy.

Datarella has joined forces with Fetch.ai and Ocean protocol over the last few months to build an Ocean Connection Agent to take private data available to itself, store it confidentially, and then spin up a fixed-rate exchange on the Ocean Market or with compute to data enabled as shown in the previous figure. You can imagine that in the near future, this might be your current location data or data about your preferences or searches performed locally within an app (the MOBIX app for example). This data never leaves the sovereign control of the user and the user doesn’t have to worry about handling or organizing or authorizing the data because the AEA does that on behalf of the user well, autonomously! The same applies to companies running AEAs on IoT devices too. You might be asking now why this is even beneficial. It’s useful because it solves the problems associated with those vast data lakes that are driving the recommendation engines behind all sorts of “smart” services.

The Solution for Automating Private Business Intelligence

Ocean Marketplace is an open-source decentralized market that you can fork to develop your own marketplace. It comes equipped with all the features right out of the box. The “Compute to Data” part enables the person, or in this case, AEA is posting the data to stipulate that it cannot be downloaded in the clear – but rather that you’re selling the results of selected algorithms which you or your agent will run (automatically) against the private data on behalf of a purchasing entity (person or AEA). They get the results of the computation – the basis for those context-aware recommendations, but they don’t have the opportunity to abuse the data because they never actually see it despite gaining actionable business intelligence based on the contents of said data. That’s what we mean by “Automating Private Business Intelligence”.

The world we’re building is one where your data remains yours but the results of computations on your data will enable a host of new context-aware smart services that respect your privacy by design but still feel like magic. Bringing the unique properties of the Ocean Marketplace together with Fetch.ai Agents means for the everyday person that it’s increasingly going to be possible that you can have your cake and eat it too. You won’t have to choose between amazing functionality and privacy – you’ll be able to have both – and get paid for it.  Your data – your profits!

Developer Resources

If you’re a developer and want to dive into the code there are two spots you’ll want to check out.  First of all, you’ll find the code open-sourced on github here. Additionally, if you’re already an AEA developer you’ll want to check out the package in the AEA registry. In the near future, we’re also planning to release the Storj AEA but for now, you can take any S3-compatible storage you choose. Let’s start automating private business intelligence generation!

The post Automating Private Business Intelligence appeared first on DATARELLA.

The path to the fourth industrial age is being paved by the interplay of Big Data-driven automation, robotics, IoT and Distributed Ledger Technologies, aka Blockchain. Given the increasing amount of data, and the number of digital services that go hand in hand with this progress, the need to automate them is also growing. Users should be relieved of unnecessary work and offered optimal results.

Agents take on the role of autonomously performing tasks on behalf of their clients (individuals or objects). For this purpose, they can also interact with each other. Intelligent agents can make complex decisions by using ML, i.e. AI-powered algorithms, based on large amounts of data.

A blockchain, with its data supply, offers a particularly favourable environment for intelligent agents. The data of a blockchain are permanently available and are logically related to each other. Decentralisation can offer robustness (no single point of failure) and lower transaction costs. Agents can assume a fully autonomous identity on a blockchain through private keys. They can use it to authenticate themselves and communicate their suitability for certain tasks. Agents can use shared protocols (possibly through smart contracts) to coordinate, collaborate efficiently, e.g. by distributing complex tasks among themselves. They can negotiate and make distributed decisions (even though voting processes use their blockchain). Tasks, goals or motives of agents can be recorded in the blockchain and economic incentives can be set for optimal task performance.

With regard to the IoT, a blockchain (as a single point of truth) can integrate various sub-systems, s.a. smart household appliances, smart buildings, smart districts and smart cities, and create added value for all agents participating in the network. Fetch.ai is an example of how intelligent agents can realize automated services based on blockchain technology.

Among the use cases of Fetch.ai, we would like to highlight agents for mobility services – traffic sign agents, parking agents for Deep Parking, agents for eMobility, agents for trains and stations that could even form a decentralised train network. In the process, increasingly intelligent autonomous agents interact on behalf of people or infrastructure, searching for each other, negotiating with each other in the interest of offering their users optimal solutions. In such a case, an autonomous agent of a car could, on behalf of its owner, seek and negotiate with agents working on behalf of parking lots to navigate the car and its owner to a quick and cheap place to park. With Deep Parking at the IAA in Munich 2021, the potential of agents for such use cases becomes clear.

There it was demonstrated how agents negotiate their resources on behalf of vehicles, their owners and the infrastructure to find an optimal solution for everyone without further efforts for the users. The following graphics show an excerpt from the exemplary communication between the agents involved.

In this case, a user named Jane is looking for available parking space in the city centre. Without Jane having to do this herself, the agent in her car (My Agent (Car)) looks for another agent who offers a parking space via a specific (agent-)network (SOEF). Using Blockchain technology, agents handle authentication, price negotiation, reservation and even payment, autonomously according to their client’s preferences. When Jane approaches the parking lot, access is automatically granted to her car without further ado.

As shown, the scope of tasks autonomous agents can perform and the added value they can contribute is without limits. So, by using autonomous agents, the potential of Blockchain technology can be leveraged for all use cases where handling of huge amounts of data in real-time or near-time is needed.

The post Autonomous Economic Agents – Automation Services for Blockchains appeared first on DATARELLA.