Case management – the invisible backbone of justice delivery – has for decades remained a largely manual, fragmented process, weighed down by paper trails, human bottlenecks, and opaque systems that feel like a Sisyphean nightmare (or perhaps the purgatory of bureaucracy – the line between the two is vanishingly thin). And yet, in 2025, we find ourselves on the cusp of a radical shift. Why? Because Artificial Intelligence (AI) – the hype, the buzzword, the enigmatic force that has swept across every industry – is finally rewriting the rules of justice. Far beyond simply digitising old processes, AI is beginning to reimagine how justice itself is delivered, promising a transformation that feels both long overdue and unstoppable.

Having spent years observing and engaging with legal technology, one can say with conviction that this change is not about automation alone – it is far more profound. Where once technology merely converted paper files into digital records and shaved time off clerical tasks, AI-powered case management now changes how information is understood, shared, and acted upon. It reshapes the flow of dialogue in courtrooms, and redefines what it truly means to have timely, fair access to justice – in ways that, just five years ago, seemed improbable, if not impossible.

The Slow yet Steady March Toward Digitalisation

It is easy to forget how recent this transformation really is. Not long ago, digital case management meant little more than electronic filing – scanned documents, basic workflow tracking, and limited online access for lawyers. These so-called innovations often replicated the inefficiencies of paper systems in a digital shell. A missing signature could still derail a case, data stayed locked in silos, and citizens rarely saw their own case progress beyond what their lawyer told them.

The pandemic shifted that trajectory. Courts went virtual, and legal professionals turned to tools they once resisted. Suddenly, there was no appetite for ‘business as usual.’ AI stepped in – not as a distant promise, but as a practical solution. Today, the question is no longer whether legal workflows will be digitised, but how intelligent they can become, and how effectively they will serve people.

AI’s Quiet Revolution in Case Management

Artificial intelligence has entered case management not with a bang, but through a series of quiet, transformative shifts. Intelligent systems now scan thousands of pages in minutes, extract key facts, spot patterns, and link related cases with speed far beyond human capacity. Predictive analytics help teams anticipate outcomes, while anomaly detection flags red flags long before they reach court.

Most profoundly, AI is changing the very pace and experience of justice. Delays once seen as inevitable are no longer excusable. Automated systems track timelines in real time, keeping filings, hearings, and judgments moving. Citizens can log into secure portals to see updates, upload documents, or receive plain-language AI summaries – simple features that signal a cultural shift in access to justice.

The Challenges of Intelligence

Yet, the age of AI in case management is not without its challenges. There are lingering fears about transparency and fairness. If an algorithm flags a case as ‘low priority’ based on historical patterns, does it inadvertently replicate systemic bias? If predictive tools suggest likely outcomes, do they influence judges and lawyers to follow the path of least resistance?

These are not theoretical concerns. They are real dilemmas, currently unfolding in jurisdictions that have adopted first-generation AI case tools. Many innovators, including those of us working alongside the visionaries at elint AI, believe that the future of digital case management hinges on explainability. Systems must not only produce results but also reveal how those results were reached. This is the foundation of trust in justice – and technology must adapt to uphold it. As U.S. Chief Justice John Roberts once cautioned, ‘Human judges will still be needed, but AI will influence the judiciary… any AI use requires caution and humility.’

Looking Ahead: A Future Worth Building

The next frontier for legal technology is not just managing cases but building true case intelligence. Picture a system that does not merely track documents but actively learns from case law, social trends, and procedural patterns to recommend smarter paths toward justice. A system that spots recurring hurdles and alerts policymakers before they calcify into systemic failures. One that can guide self-represented litigants through complex processes with the care and precision of a seasoned lawyer.

This is not wishful thinking. Platforms like Lex Machina, Theo AI, Justice Accelerator, and Bench IQ are already pushing toward this vision, blending advanced natural language processing, real-time analytics, and predictive reasoning to transform case management from a static record-keeping task into a proactive force for fairness.

The technology is here, but the mindset shift is harder. AI challenges long-held habits – the comfort of paper trails, the notion that slowness equals diligence, the belief that only humans can meaningfully ‘understand’ a case. Yet clinging to tradition keeps justice slow, soulless, and out of reach of those who need it most.

AI is not a threat to justice; it is a chance to make its long-promised ideals real – faster timelines, clearer processes, fairer outcomes. The future of case management is intelligence in service of one principle that must never change: justice for all.

The post Digital Case Management in the Age of AI: Rethinking Justice for a Connected Era appeared first on Enterprise Viewpoint.

This shift became clear to me during a break from the storage industry after 15 years. I had returned to school at Stanford, intending to join an early-stage company, and did not expect to jump back into storage. That changed when my co-founder showed me a prototype he’d built of a cloud file system that behaved in a way nothing like the traditional models I was used to. When accessing files stored in the cloud, it didn’t rely on creating replicas, didn’t require syncs or long download times, and ran directly on top of cloud object storage. It felt like a workaround for a problem I hadn’t realized was solvable.

That prototype eventually became LucidLink, but the larger takeaway was this: the future of file storage isn’t about simply moving data to the cloud. It’s about changing how we store and access data in the first place.

File Systems Weren’t Built for This

Most file systems, even cloud-based ones, are still rooted in assumptions that made sense when work was co-located and bandwidth was a limiting factor. But today, teams are distributed, files are huge, and “real-time” collaboration is often taken literally. Many teams, especially those in creative industries like media, architecture, and design, are juggling multi-gigabyte assets across time zones and toolsets.

Traditional solutions can’t always keep up. VPNs are slow, and while bandwidth has become ubiquitous, latency cannot be changed. Sync-and-share platforms often create versioning headaches. Shipping hard drives by mail is still a thing. The result is friction and inefficiencies that are now baked into the daily experience of trying to collaborate. This doesn’t just slow down projects; it actively impacts productivity, stifles creativity, and ultimately, compromises an organization’s competitive agility.

Why Object Storage Matters

Object storage has become the default for scalable, durable cloud infrastructure. Its design—flexible, redundant, and cost-effective—makes it ideal for modern IT environments. But on its own, object storage wasn’t built to serve as an interactive file system. It excels at storing large volumes of data, but not at delivering low-latency access to individual files across continents.

This is where we’re seeing innovation: layering intelligent software on top of object storage to deliver a fast, real-time file system experience, making cloud data feel local and instantly accessible to users, without giving up the benefits of scale and resilience. This model doesn’t just support cloud workflows; it enables new ones.

Several providers are now exploring or delivering versions of this architecture, including integrations with IBM Cloud Object Storage. These solutions aim to give users the experience of a local drive, while pulling data directly from object storage. They stream only what’s needed and when it’s needed rather than downloading or uploading entire projects in order to work on just one small part.

Use Cases That Make It Real

One example I often come back to is Torti Gallas + Partners, an architecture and planning firm working across multiple offices and time zones. They were dealing with version confusion, slow file access, and storage systems that didn’t scale well with their distributed team. By moving their project files to a shared cloud environment layered over object storage, they were able to consolidate workflows and reduce the friction of collaboration. Importantly, they didn’t have to overhaul their toolset or retrain staff to make it work.

What’s notable isn’t that this solved a problem, it’s that this kind of setup is becoming increasingly viable. Teams can now collaborate across borders without spending days syncing files, managing version conflicts, or relying on fragile handoffs.

The storage conversation is no longer about capacity. It’s about workflow. As the lines blur between storage, collaboration, and application performance, file systems need to meet users where they are. Object storage, once seen as cold and passive, is becoming the active backbone of those workflows. With the right architecture and partners who truly understand this paradigm shift, it can support not just where your data lives, but how teams use it. And that’s what matters — a fundamental change that is defining the competitive advantage for data-driven enterprises today and tomorrow.

The post Why It’s Time to Rethink How Enterprises Store and Access Data appeared first on Enterprise Viewpoint.

Construction still accounts for 21% of all workplace fatalities in the U.S., a staggering statistic for any sector. And yet, most people in this industry, whether in the field or in the office, care deeply about safety. We look out for our crews, mentor new hires, and take pride in getting the job done right. At its core, construction is built on teamwork, accountability, and trust. However, as projects have grown more complex and safety regulations more layered, the way we implement safety has begun to drift from how people actually work. The systems in place, designed to reduce incidents and improve accountability, often feel like administrative burdens rather than tools that support real safety.

This disconnect doesn’t stem from apathy among workers. The challenge lies in how safety is delivered. The systems we’ve built don’t always reflect the way people in construction learn, think, or operate. I’ve seen this evolution firsthand. I started my career in construction over twenty years ago as a laborer. Back then, safety was mostly about the basics: wear your hard hat, steel-toed boots, and reflective vest. There were few formal checklists, no extensive manuals, and little in the way of regulated programs. If someone got hurt, it was often accepted as part of the job—an unfortunate but expected risk. We learned by watching, by doing, and sometimes, by learning the hard way.

I eventually started my own construction company, installing windows on high-rises—a high-risk job. Like many others, we had good intentions but little structure. That changed when a general contractor refused to let us on-site without a formal safety program. For the first time, we had to document our processes, show our training records, and submit daily and weekly forms. And suddenly, I was trying to get a crew of 50-year-old glaziers to fill out paperwork they hadn’t touched in a decade. At first, it felt like just another hoop to jump through. But then something happened that changed everything.

We were working on a building when a steel angle fell from the 14th floor into a Whole Foods parking lot. Fortunately, no one was hurt, but it could have been a tragedy. I’ll never forget the fear in my foreman’s voice as he called me, running down the stairs, thinking someone might have died.

In that moment, the conversation shifted for me. Safety wasn’t about avoiding fines or passing inspections. It was about people. It was about the weight of responsibility I carried for every worker under my leadership. And I realized that the systems we had in place, no matter how well-intentioned, were too reactive.

Why “Checking the Box” Isn’t Enough

What we’ve built in many parts of the industry is a version of safety that centers on compliance: “Fill out this form.” “Complete that checklist.” “Take this training.” But often, we don’t ask the bigger question: Is this actually making the work safer?

In our effort to formalize safety, we’ve sometimes created a disconnect between the rules and the people they’re meant to protect. Field teams often see safety as something that slows them down. It’s a hurdle, not a tool. But safety should be intuitive, just like picking up a hammer or operating a lift. We don’t remind workers to use their tools; they just do it. That’s what safety should feel like. It should be part of the rhythm of the work, not something extra layered on top of it.

There’s also a misconception that safety is natural, that people will always make the safe choice. But that’s not how we’re wired; we’re human. We prioritize speed. We get comfortable. And when that happens, we become complacent.

Complacency is dangerous because it’s invisible. A worker might stop tying off because “nothing’s ever happened before.” But the risk hasn’t changed, just our perception of it. The real challenge is building a culture that interrupts complacency. And that doesn’t come from policies or penalties. It comes from creating space for people to connect with why safety matters—to them, to their families, to their crew.

A Field-First Approach

After that near-miss at the jobsite, I began to rethink how safety could work, not just for my company, but for the industry. I co-founded SALUS to solve a very specific problem: how do we make safety something that feels useful, not burdensome?

Our solution wasn’t just to digitize forms. It was to design a system that workers would actually want to use. That meant making it mobile-friendly, reducing complexity, and delivering value directly to the field. Because when information is accessible, when processes are intuitive, and when technology fits into the flow of work, not the other way around, people engage. And once workers are engaged, safety professionals can move beyond just collecting forms. They can start driving real improvements, focusing on behavior, mindset, and communication.

One of the most critical shifts we need to make is how we measure safety. Right now, our industry relies on lagging indicators like EMR (Experience Modification Rate) or TRIF (Total Recordable Incident Frequency). These numbers tell you how many people got hurt, but not why. And worse, they create incentives to underreport incidents instead of learning from them.

Imagine if we treated safety incidents the way the airline industry treats crashes: as lessons for the entire field. In aviation, every major incident is studied worldwide. In construction, we often keep those lessons locked away—or worse, ignore them entirely.

We need to remove the stigma around reporting. Near misses should be celebrated as opportunities to prevent harm. Instead of holding companies responsible for admitting faults, we should reward those who are transparent and committed to continuous improvement.

So how do we get there? We start by breaking down silos. Safety isn’t just the responsibility of a foreman or a safety manager. It’s owned by everyone, from laborers to executives. And that only happens when trust is built, the systems in place make sense, and every worker understands that their voice matters.

We also need to invest in leadership, specifically, safety leadership. Not just people who enforce rules, but people who inspire change. We need to treat safety professionals like CFOs: strategic leaders, not administrators. And above all, we need to stop thinking of safety as a box to check and start treating it as a core part of operating intuitively, intentionally, and with care.

The systems we’ve built got us this far, but they won’t take us where we need to go. We’ve flatlined. And if we want to move forward, we need to stop drowning in paperwork and start listening to the people who do the work daily.

Safety isn’t just about forms, inspections, or regulations. It’s about people. It’s about progress. And it’s about building a culture where the value of a life outweighs the pressure of a deadline.We have the tools. We have the stories. Now we need the will to change.

Because at the end of the day, what matters isn’t how many forms were filled out—it’s whether everyone made it home safely.

The post The Future of Construction Safety Is Cultural, Not Procedural appeared first on Enterprise Viewpoint.

Early in my career, I was that planner desperately trying to wrangle data from multiple reporting systems, copying and pasting into Excel, and hoping my formulas were correct. The manual processes were not only inefficient but created countless opportunities for error. Sound familiar? I spent countless hours each month reconciling data sources, rebuilding broken links, and explaining why my numbers didn’t match finance’s numbers or operations’ numbers.

Then I encountered IBM Planning Analytics at an organization that had implemented it for enterprise planning and forecasting. The transformation was immediate and profound. Instead of wrestling with disconnected spreadsheets, we leveraged the powerful Excel add-in to build collaborative planning models. Data flowed seamlessly across departments, monthly projections updated efficiently, and for the first time, everyone was literally working from the same numbers.

The AI Revolution: Learning from Early Missteps

Since those early Planning Analytics days, the landscape has evolved dramatically, particularly with the emergence of generative AI. In my current role, I regularly speak with companies eager to leverage advanced analytics and automation. However, I also hear familiar stories of AI implementations gone wrong.

I’ve been there myself. At one organization, we purchased an AI forecasting tool that promised to revolutionize our planning process. Instead, it became a source of frustration. The “black box” approach meant we couldn’t understand how the system arrived at its recommendations. My team and I found ourselves constantly overriding the machine’s output, ultimately trusting our manual calculations more than the automated results.
This experience isn’t unique. When I discuss AI solutions with business professionals today, I encounter significant skepticism. They want to understand the methodology behind any recommendation. They’ve been burned by tools that promised intelligence but delivered opacity. The good news? The technology has fundamentally changed.

IBM watsonx: Transparent Intelligence for Business Planning

Today’s AI platforms, particularly IBM watsonx, address the transparency concerns that plagued earlier implementations. IBM watsonx provides a comprehensive environment for housing and scaling AI models while maintaining visibility, and is built with IBM security. IBM is a trusted enterprise solution provider and an essential part of the AI strategy within many large organizations.

Consider the possibilities: your organization possesses years of historical data containing patterns that human analysis might miss. Machine learning excels at identifying these patterns and applying them to future predictions. Layer in external data sources like weather patterns, consumer price indices, or market indicators, and you create forecasting capabilities that far exceed traditional approaches. The machine provides consistency and fairness in pattern recognition, allowing your team to focus on strategic decision-making rather than data manipulation. Current applications we’re exploring with clients include:

• Natural language model interrogation (similar to ChatGPT functionality)
• Automated presentation generation
• Scenario modeling and operational predictions
• Forecast automation with human oversight
• Intelligent variance analysis
• Missing data imputation
• Correlation identification
• Executive reporting automation

Planning Analytics: The Foundation for Advanced Analytics

Not every organization is ready for AI implementation, and that’s perfectly acceptable. Many companies still rely heavily on offline Excel for critical planning processes. IBM Planning Analytics provides an excellent stepping stone toward advanced analytics maturity.

This purpose-built planning and forecasting platform serves organizations ranging from Fortune 100 enterprises to mid-market companies with 10-15 users. Users can access the system through the familiar Excel add-in or web-based visualizations, depending on their preferences and requirements.

Planning Analytics eliminates the data silos inherent in spreadsheet-based planning. Multiple users collaborate seamlessly across the organization, conduct scenario analysis, maintain version control, and establish official forecasts. Most importantly, everyone operates from a single source of truth, backed by IBM’s enterprise-grade security and infrastructure.

The platform’s customizability sets it apart from one-size-fits-all solutions. Whether deployed on-premises or in the cloud (IBM, AWS, or Azure), Planning Analytics adapts to your business complexity rather than forcing you to adapt to software limitations. This flexibility drives higher user adoption rates and delivers measurable business value.

Moving Forward on Your Analytics Journey

Every organization occupies a different position on the analytics maturity spectrum. The key isn’t where you start, but that you start moving forward. Whether you’re transitioning from Excel to Planning Analytics or exploring AI-powered forecasting with watsonx, the important thing is taking that next step.

The technology landscape will continue evolving, and organizations that embrace this position and continue on their analytics journey will fine themselves having a competitive advantage. Your analytical journey doesn’t require a destination, just forward momentum and the willingness to leverage better tools as they become available.

The future of business planning is here. The question isn’t whether you’ll eventually adopt these technologies, but how quickly you’ll realize their benefits.

The post Where are you on your Analytical Journey? appeared first on Enterprise Viewpoint.

This is where today’s infrastructure creates a major obstacle. Edge data is highly distributed across disparate locations, stored on low-end servers and gateways, and varies in format and frequency. There are no standardized data services at the edge, meaning organizations are forced to centralize the data in the cloud before using it. This results in latency, increased cost, and reduced reliability—precisely the opposite of what agentic AI requires.

The convergence of decentralized data infrastructure and intelligent edge orchestration is now reshaping this reality. AnyLog eliminates the need for centralization by creating a decentralized, queryable, and secure data layer directly across edge nodes. It allows AI agents to access and act on real-time data where it is generated. When combined with IBM Edge Application Manager (IEAM)—which provides autonomous deployment and orchestration of software across thousands of edge devices—this integrated solution enables scalable, resilient, and secure deployment of agentic AI. Together, AnyLog and IEAM deliver a plug-and-play foundation for autonomous operations at the edge.

AnyLog: Empowering Autonomous Agents with Decentralized Data Access

AnyLog addresses the challenges of data centralization by virtualizing edge infrastructure into a real-time, queryable, and self-managed network. This approach allows AI agents to access and act upon data directly at the edge, eliminating the need for data to traverse to centralized cloud systems. (see details on Medium)

Key features of AnyLog include:

• In-Place Data Querying: Agents can access distributed data without relocating it, reducing latency and preserving data sovereignty.
  
• Localized Computation: Processing occurs near the data source, enhancing resilience and efficiency.
  
• Decentralized Connectivity: Agents connect through a secure, decentralized overlay, eliminating dependencies on static configurations.
  
• Self-Monitoring Nodes: Nodes autonomously report their state, facilitating automatic agent discovery and coordination.
  

By transforming the edge into a network-aware, cryptographically secured data layer, AnyLog provides the foundational infrastructure for scalable and secure agentic AI.

IBM Edge Application Manager: Orchestrating Workloads at Scale

IBM Edge Application Manager (IEAM) complements AnyLog by providing a robust platform for managing and deploying workloads across a vast network of edge devices. Built on Open Horizon open-source software, IEAM enables autonomous management of edge computing environments.

Highlights of IEAM include:

• Centralized Management Hub: Facilitates the deployment and monitoring of workloads from a central location to remote edge nodes.
  
• Policy-Based Deployment: Administrators can define deployment policies that govern where and how services are deployed, ensuring compliance and efficiency.
  
• Autonomous Lifecycle Management: Edge nodes can independently manage the software lifecycle, including updates and monitoring, reducing the need for on-site IT personnel.
  

IEAM’s architecture is designed to support remote operations of edge computing facilities, making it ideal for industries with distributed operations such as manufacturing, retail, and logistics. 

Synergizing AnyLog and IEAM for Agentic AI

The integration of AnyLog and IEAM creates a synergistic environment where agentic AI can thrive.

• Data Accessibility Meets Orchestration: While AnyLog ensures that AI agents have immediate access to decentralized data, IEAM provides the tools to deploy and manage these agents across the edge network efficiently.
• Enhanced Autonomy: Agents can make informed decisions using real-time data from AnyLog and execute tasks seamlessly through IEAM’s orchestrated environment.
• Scalability and Resilience: The combined solution supports the deployment of thousands of agents across diverse locations, each capable of autonomous operation and collaboration.

This integrated approach is particularly beneficial for applications requiring real-time decision-making and action, such as predictive maintenance, autonomous vehicles, and dynamic supply chain management.

Conclusion

The fusion of AnyLog’s decentralized data platform with IBM Edge Application Manager’s orchestration capabilities paves the way for the next generation of agentic AI applications. By enabling autonomous agents to access and act upon data in real time across distributed environments, organizations can achieve unprecedented levels of efficiency, responsiveness, and innovation.

The opportunity is transformative: industries ranging from manufacturing and energy to logistics and smart cities can replace manual monitoring, siloed systems, and slow decision cycles with real-time, autonomous operations. Agentic AI powered by edge-native infrastructure allows enterprises to operate faster, more intelligently, and at massive scale—without relying on the cloud or building bespoke solutions for every edge deployment.

Beyond the performance gains, the cost savings are substantial. By automating decision-making and control directly at the edge, companies reduce the need for human intervention, lower operational overhead, and minimize costly downtime. Additionally, reducing the dependency on centralized cloud services significantly cuts bandwidth costs, eliminates the need for large-scale data transfers, and enhances data privacy and sovereignty.

This shift not only redefines what’s technically possible at the edge—it also reshapes the economics of digital transformation, unlocking new business models while making edge AI deployments scalable, secure, and sustainable.

The post From Cloud Limits to Edge Intelligence Powering Agentic AI with AnyLog + IBM IEAM appeared first on Enterprise Viewpoint.

However, these models possess a characteristic that directly conflicts with the traditional programming paradigm: their output is non-deterministic, which is fundamentally at odds with conventional code. When we cannot know with certainty what format we will receive, it becomes difficult to chain tasks, invoke APIs, extract or persist information, or handle other functionalities programmatically.

Consider some use cases where obtaining a predictable response can be vital:

• Automatically evaluating customer or user reviews of my products according to specific categories, storing the classification in a database table, and generating a BI dashboard or alerts for products with high dissatisfaction.
• Extracting specific fields from legal or tax documents (dates, amounts, counterparties) without relying on the style in which the text was written.
• Evaluate responses from a customer service AI chatbot in terms of helpfulness, kindness, hallucination or other metric.
• Routing user query to a specialized agent based on the content of that query.

All these problems share the fact that they involve free-form text input that can be very difficult or impossible to manage with regular expressions or NLP techniques, and that they require a fixed output. The solution: structured outputs.

Structured outputs are a capability that leverages the function-calling feature offered by major LLM providers in some fine-tuned models, and which IBM watsonx exposes natively, to deliver responses in a predetermined JSON format. You can think of this capability as the developer writing a contract (a JSON schema) that the model reliably fulfills. In watsonx, models such as Granite 3.3 8B, Llama 4 Maverick and others support structured outputs thanks to function calling. This opens the door to countless applications, from simple information extraction or classification solutions to chained tasks where one model’s output becomes the next model’s input, like an agent pipeline, or feeds a conventional function within a code pipeline. The result is more maintainable code and often allows you to distribute the cognitive load of much more complex processes (divide and conquer).

Example

To better illustrate how practical this can be, let us examine a simple example in a Python notebook using:

• Granite 3.3 8B as the LLM in IBM watsonx.
• Langchain, a library that facilitates the use of language models.
• Pydantic to define and validate the structure we expect.

The process we want to automate is the extraction of information from invoices. We have the invoice content, but each invoice is different, with formats that vary greatly in how they present the information. Below is one example format among the many we might encounter:

From this, we need to extract:

• Purchase order number
• Invoice date
• Subtotal amount
• Tax amount
• Total amount
• List of products (an indeterminate number of items, for which we need name, quantity, and unit price)

First, we will install the appropriate libraries and import them into our notebook where we will perform the exercise. We must have the necessary credentials to connect to watsonx.

We can create a simple prompt that helps the model understand its role, the problem it must solve, and guides it through potential issues it may encounter. Below is an example in which we insert the invoice content:

Once we have a basic prompt, we need to generate the structure we will request from the LLM. For this, we will use Pydantic. In this case, we will define the schema for each individual product and the invoice schema, which will include a list of products.

Then we instantiate our LLM by specifying the model ID we want to use, in this case Granite, and the project ID.

Finally, using the .with_structured_output method, we provide the expected output schema and invoke the model with the prompt that includes the invoice information we are analyzing.

Upon obtaining the result, we can see that the model perfectly adhered to the requested format based on the data we provided and did so in a very short time. Furthermore, behind the scenes it validated the response against the schema, so we do not need to implement complex logic for this.

This is a very simple example; however, we can extend it much further. Perhaps we need the date in a specific format or want to provide instructions at the level of each schema field. Or maybe we want to use this model’s response to feed an internal system or another model with a task that requires the invoice data to be clean.

Link to sample code: https://github.com/matiasBarrera98/structured_outputs

Summary

The ability to generate structured outputs in IBM watsonx bridges the gap between the creativity of the models and the predictability demanded by traditional code. It delivers reliable responses that can be chained into subsequent tasks without complex validation, cleaning, or advanced NLP techniques, turning LLMs into components ready to integrate into any enterprise pipeline.

I encourage development teams and AI professionals to experiment with this functionality and take advantage of the wide range of possibilities it offers as an additional tool. With watsonx and structured outputs, software development and generative AI converge like never before, providing a competitive advantage worth exploring starting today.

The post Structured outputs with IBM WatsonX appeared first on Enterprise Viewpoint.

Proposed as a theory in the 1980s, quantum computing envisions a revolutionary approach to the construction and operation of computers. Classical computing executes deterministically on data and instructions represented as binary digits or bits. Quantum computers define quantum bits, or qubits, using the quantum characteristics of superpositioning, entanglement and uncertainty—offering a much faster, probabilistic approach to solving complex mathematical problems. The path from theory to reality, however, has been slow. For the past 40 years, meaningful quantum computers have always been said to be five to ten years off.

2025 marks a turning point where practical, useful quantum computing goes mainstream, and IBM is leading the way. This article describes IBM’s roadmap, results achieved and use cases moving from the lab into the real world.

The road more travelled
IBM’s quantum computing roadmap dates back nearly a decade, to early experimental work in 2016. Like many others, the company built prototype quantum computers out of only a handful of qubits but also provided access to its hardware over the cloud and fostered a robust developer ecosystem centred on its Qiskit tooling.

In the early 2020s IBM dramatically scaled up its quantum hardware, beginning with the 127-qubit System 1 (Eagle) in 2021, and continuing with the 433-qubit Osprey in 2022, and the 1,121-qubit Condor in 2023. The System 1 is now the foundational workhorse for the IBM Quantum Network, providing customers direct access to test and execute experimental quantum workloads. It has also been installed standalone in Canada, the United States, Germany, Japan and other countries.

IBM updates its roadmap annually, marking completed milestones and displaying ambitious targets into the next decade. Hardware and software mature together, including enhancements to Qiskit, and supporting complex workloads with circuit depth increasing exponentially from a current 5,000 gates. A gate is roughly equivalent to a single quantum machine instruction, so it’s a good measurement of the complexity of quantum algorithms.

Bigger isn’t better
The largest problem afflicting quantum computing is high error rates due to qubit decoherence. One cause of decoherence is interference from other qubits, effectively limiting the vertical scalability of quantum hardware. As a result, IBM’s Condor proved to be more of an engineering proof of concept, and other techniques were required to grow quantum computing capacity.

IBM’s focus shifted to developing parallel processing in quantum computing, based on the 133-qubit Heron introduced in late 2023. Subsequently, IBM announced and deployed a Quantum System 2 in its TJ Watson Research Center in Yorktown, New York. The System 2 currently deploys three Heron processors and additional work is being done to stabilize couplings, adding more processors. The Qiskit tools have also been improved to enable circuit cutting and circuit knitting, distributing the computational workload across processors.

Expect quantum computers to follow this pattern and expand horizontally, not vertically, in the years to come.

Dealing with errors
Quantum computers are notoriously error-prone, with an error rate of about one in a thousand, compared to a classical error rate of roughly one in a quintillion (10¹⁸.) This is the biggest inhibitor to achieving quantum advantage, the point at which quantum computers will meaningfully outperform classical computers.

In 2023, IBM introduced error mitigation—mathematical techniques to reduce the effect of errors. Zero-noise extrapolation, for example, uses detectable patterns in quantum errors to extrapolate a better approximate solution and can be included in quantum circuits via a simple include statement using Qiskit. Error mitigation is a good intermediate step in improving the quality of quantum computers but adds overhead.

IBM’s goal is fault-tolerant quantum computing, by devising logical qubits out of multiple physical qubits. This is tricky to achieve, and current surface codes, or connections between physical qubits, aren’t robust enough—requiring hundreds or thousands of physical qubits for a single logical qubit. In early 2024, IBM released a new surface code topology, arranging qubits in a virtual torus shape resulting in a ten-fold reduction in the number of physical qubits required for a logical qubit. A proof of concept was demonstrated showing 12 logical qubits formed from 288 physical qubits, where nearly 3,000 would have been required previously. More work will be needed to scale this up to hundreds of logical qubits, but it’s a step in the right direction.

With error mitigation and early steps toward fault tolerance, IBM has achieved quantum utility—an intermediate step before quantum advantage, where quantum computers can produce results at least equivalent to high-performance classical computers, and the two can validate each other. The company has published several practical use cases in condensed-matter physics, statistics and materials science that demonstrate quantum utility.

A look ahead
IBM has consistently met the key milestones in its roadmap through 2024, and the next few years will bring further improvements in scalability and fault-tolerance. The Heron processor will grow from 133 to 156 qubits this year, and the System 2 will expand to seven parallel Herons. By 2029, this will lead to an improvement in quantum circuit quality up to 100 million gates. By then, expect IBM to have achieved quantum advantage with relevant use cases deployed in business and science.

Quantum, however, is not a general-purpose computing solution and it won’t replace classical computing. The future will be hybrid, where quantum and classical computers combine forces to solve the most complex problems which neither can solve alone. Watch for IBM to elaborate its vision of quantum-centric supercomputing before the end of this decade.

My main ask of IBM is, get out of the lab! Quantum has always been a research activity for the company, but quantum utility is good enough for customers to invest in production today.

Let’s commercialize quantum computing now. It’s ready.

The post Out of the lab and into the office appeared first on Enterprise Viewpoint.

The Challenge: Making Fan Engagement Scalable and Personalized

Traditionally, engaging with fans at scale has been limited by venue capacity, broadcast boundaries, and lack of personalized experiences. With the digital-native Generation Alpha and Gen Z becoming influential audience segments, the demand for intelligent, emotionally resonant, and interactive sports content is at an all-time high.

SpeedShotX addresses this demand by merging AI, computer vision, and augmented reality to create dynamic and personalized fan experiences. Whether fans are at the stadium or watching from home, the technology offers real-time insights, stats, interactive content, and brand tie-ins that respond to the context of the game.

The Role of watsonx

IBM’s watsonx platform has been central to enabling this breakthrough. The modular architecture, governance frameworks, and trusted data capabilities provided by watsonx.data and watsonx.governance allow us to:

• Train domain-specific AI models for understanding live sports feeds and fan sentiment.
• Ensure data privacy and compliance at scale for global sports applications.
• Deliver responses in sub-second timeframes—critical for real-time fan experiences.
• Generate meaningful, personalized content through generative AI features.

Using watsonx, we have created a robust foundation for AI models that don’t just process data, but make decisions, adapt, and learn based on live fan engagement patterns.

SpeedShotX in Action          

Engage with the F1 Activation

SpeedShotX is designed to be mobile browser-based, language-agnostic, and sport-agnostic—capable of operating seamlessly on any phone, in any language, anywhere in the world. This inherent flexibility positions SpeedShotX to reach over 3 billion sports fans globally, delivering personalized and immersive experiences without requiring app downloads or device-specific customizations.

The sports industry represents a multi-trillion-dollar global economy, encompassing everything from media rights and sponsorships to merchandise, ticketing, fitness, gaming, and fan-driven commerce. In 2024, the global value of sports media rights surpassed $62.6 billion, fueled by flagship events such as the Summer Olympics and UEFA European Championships. Meanwhile, the global sports sponsorship market reached $76.3 billion and is projected to grow at a CAGR of 8.7%, potentially doubling by 2034 to exceed $160 billion. This ongoing growth underscores the immense potential for scalable technology platforms to meet the evolving demands of fans, brands, and rights holders alike. SpeedShotX offers a turnkey solution designed to tap into this vast market, enabling brands to connect with over 3 billion fans through personalized, culturally relevant, and AI-powered real-time interactions delivered anywhere in the world. SpeedShotX captures and analyzes images from live games using mobile devices. Within milliseconds, it returns context-aware information:

• Player identification and stats based on jersey numbers and facial recognition.
• Historical data and highlight reels curated to match the ongoing moment.
• Brand activations personalized to each fan’s preferences and past interactions.
• Merchandise and ticketing opportunities linked contextually to the game experience.

All of this is processed through an AI engine that learns with each interaction, evolving to offer smarter, richer experiences as engagement grows.

Content Delivery at Scale with IBM COS + CDN Integration

To support this AI-powered, immersive experience, a robust content delivery architecture is crucial. Fiducia | AI has adopted IBM Cloud Object Storage (COS) integrated with global CDNs—specifically Akamai and Fastly—to ensure fast, reliable, and scalable delivery of dynamic content including images and 3D models.

IBM COS serves as the primary origin for storing static and rich media assets. Akamai, via IBM Cloud Internet Services (CIS), offers enterprise-grade SLA-backed global edge delivery, ideal for large-scale brand activations. In contrast, Fastly brings an agile, developer-centric experience with advanced edge compute capabilities using VCL and WASM, making it well-suited for fast-moving campaigns and rapid iteration.

This modular architecture enables:

• Low-latency, high-performance delivery regardless of user location.
• Scalability from large enterprise engagements to lean, budget-conscious rollouts.
• Edge logic support for security, personalization, and custom routing.
• Integration with developer workflows, including CI/CD, CLI, and APIs.

Through our Fiducia Admin Tool, we have introduced automation to manage COS uploads, and deploy caching strategies—empowering both marketers and developers to execute seamless, global campaigns.

A Broader Vision for Sports AI

Our goal is to not just keep pace with the demands of modern sports audiences but to anticipate them. By combining watsonx with immersive technologies and a resilient global delivery framework, we are making fan engagement more human—emotional, interactive, and globally accessible. This vision also opens the door for NIL (Name, Image, Likeness) monetization, brand partnerships, and dynamic second-screen experiences that generate measurable impact.

The post Built with IBM watsonx: The New Era of Personalized Fan Engagement – At the Venue or Anywhere in the World appeared first on Enterprise Viewpoint.

Read further to learn how IBM MQ supports these modern payment rails, the technical and operational benefits a messaging infrastructure provides in financial services.  Though other payment rails exist that use IBM MQ, such as Zelle or FedNOW, this article’s focus is Fedwire and RTP.

Understanding Modern Payment Rails

The term “Payment Rails” represents the underlying systems and infrastructure used to transfer funds between entities (participants).  In the United States, the most prominent USD payment rails include, but not limited to:

• Fedwire Funds Service (Fedwire): Operated by the Federal Reserve Banks, Fedwire is a real-time gross settlement system used for high-value, time-critical payments.  It offers immediate finality and is often used for bank-to-bank transfers, securities transactions, and large corporate payments.
• Real-Time Payments (RTP): Launched by The Clearing House in 2017, RTP is the first new payment rail in the U.S. in over 40 years.  RTP is gaining adoption due to its speed, lower cost compared to traditional wire transfers.

Fedwire and RTP represent the next evolution of payments innovation for instant transfer of funds and confirmation. Financial institutions, big and small, are offering either or both services to their customers to stay competitive.  That said, these Payment Rails demand a robust integration layer to handle secure, high-volume, real-time messaging; and that’s where IBM MQ comes in.

Image Credit: IBM 

What Is IBM MQ?

IBM MQ (Message Queue) is widely regarded as the gold standard for application messaging, particularly within the financial industry.  It provides secure, reliable, and asynchronous communication between applications across various platforms and environments.  IBM MQ is trusted by global banks and financial service providers for a reason:

• High Performance: Capable of executing millions of transactions per second without compromising integrity and assures exactly-once message delivery.
• Security: Supports SSL and TLS encryption for in-flight or at rest transactions.
• Cross-Platform: Functions across traditional on-premises systems (I.E.: Linux, AIX, Windows, z/OS, IBM i) and modern cloud configurations utilizing Kubernetes (I.E.: Amazon EKS, Azure AKS, Google GKE) or even as a physical appliance!
• Scalability and Resilience: Supports clustered, high-availability deployments for mission-critical operations.

IBM MQ is essential when you need assurance that every message, representing a financial transaction, securely arrives once and once only, in the correct order.

The MQ Advantage for Real-Time Payments

The RTP network is built upon the concept of MQ request/reply messaging pattern.  For example, a participant sends a request message following the ISO 20022 standard.  These request/reply messages have a short lifespan, and if the message isn’t consumed by the expiry time set, an MQ Expiry Report is generated on the queue property and returned to the participant’s response queue.

The overall lifespan of the “transaction” is around 15 seconds, but the “hops” between the participant to RTP to participant and back are 2 seconds each between its respective queue manager.  The Clearing House’s RTP documentation provides a chart to explain message expiry further, depending on transaction type.

The RTP queue managers (right side) follow an Active-Active availability, meaning The Clearing House’s two queue managers are always operational.  The Participant queue manager (left side) can also be Active-Active or Active-Standy, per datacenter or region.

Though not shown above, Participants use a software defined (logical) network router instead of a physical one.

Finally, the queue managers (Participant and RTP) utilize SSL-TLS certificates from a Certificate Authority (CA) such as Entrust, GoDaddy, etc.  These details, as well for the MQ Objects (Queue Names, Channel Names) are part of the on-boarding process when the Participant institution fills out the necessary TCH forms.

Supporting Fedwire with IBM MQ

As of this writing, Fedwire uses a proprietary message format called Fedwire Application Interface Manual (FAIM) for sending and receiving messages with Participants, though this format is being phased out in favor of the ISO 20022 standard.

In contrast to RTP, Fedwire messages don’t require expiry to be set.  The expectation is that messages persist until they are successfully delivered and processed.  Another intriguing difference is the usage of “TEST” queues, even in production; this is in place for diagnosing connectivity issues outside of the core production queues.  A final difference to point out is the usage of two receiver channels to Participants.  One channel is meant for Fedwire transactions and the other are for account statements.

Though not shown above, Participants use physical network routers meant for two geographically separate data centers.   Participants are required to have a DR (or sometimes called a Contingency Site) and validation is performed at least annually with the Fedwire staff.

Finally, the queue managers (Participant and Fedwire) utilize SSL-TLS certificates from the Federal Reserve Banks where they are the CA, interestingly enough.  The MQ Objects (Queue Names, Channel Names) are delivered as a series of scripts, called MQSC (Script Commands), including simple shell scripts for unit testing, as part of the on-boarding process when the Participant institution fills out the necessary Fedwire forms.

Looking Ahead

As real-time/high-value payment systems evolve, the role of reliable, secure messaging platforms like IBM MQ becomes even more critical.  It’s not just about moving data from point A to point B—it’s about doing so in a way that’s elegant, secure and reliable.

IBM MQ gives financial institutions the ability to keep up with emerging payment methods while maintaining trust and compliance.  Whether integrating with The Clearing House’s RTP network or the Federal Reserve’s Fedwire service or any other integration needs, MQ ensures that every transaction meets the high standards expected in the digital financial world.

The future may bring changes in how connections are made and payments are initiated, but one thing remains clear: IBM MQ is here to stay!

The post The Crucial Role of IBM MQ in Fedwire and Real-Time Payments appeared first on Enterprise Viewpoint.

This marks a structural shift in how businesses conceive processes, assign tasks, and ultimately, how work itself is designed.

From Generative Models to Intelligent Agents

Until recently, generative AI use cases were mostly limited to narrow tasks: writing emails, summarizing meetings, generating boilerplate code, or creating reports. But as noted by IBM in The Cognitive Leap (2024), standalone models lack the ability to act with context, reason through steps, and connect with enterprise tools. This is where agents come in.

An AI agent is an autonomous entity capable of perceiving its environment (digital or physical), making decisions, planning actions, and executing them toward a defined goal. When multiple agents operate as a network, new capabilities emerge: they collaborate, coordinate, learn from the environment, and orchestrate workflows with a precision that can rival — and sometimes surpass — human performance.

Our own experience implementing agents with IBM WatsonX Orchestrate in areas like IT support, employee onboarding, customer service, accounting reconciliations, and commercial training is mindblowing: 60–80% time reduction, fewer errors, and greater adaptability to environmental changes.

A New Productivity Paradigm

These agents don’t just perform tasks — they understand context. As outlined in the Navigating the AI Frontier (WEF, 2024), agents integrate memory, planning, digital sensors, and tools to act as true autonomous collaborators. Their impact is significant: rather than improving isolated tasks, they redesign entire workflows.

Take, for example, a traditional financial audit process, which may take weeks, involve several people, manual reconciliations, and document reviews. With agents, the same process can be broken down into autonomous units (data extraction, validation, comparison, reporting), working in parallel and refining results as they go.

The gain isn’t just speed — it’s operational intelligence, scalability, and adaptive flexibility without proportional cost increases.

And What About Employment?

We can’t ignore the workforce impact. According to the Future of Jobs Report 2025 (WEF), between 2025 and 2030, 170 million jobs will be created, while 92 million will be displaced due to task obsolescence — a net gain of 7%, but highly disruptive.

Notably, 40% of employers plan to reduce staff due to automation, while 85% will invest in reskilling. It’s not mass replacement, but rather a deep reshuffling of human capital toward strategic, creative, and supervisory roles.

We’re already seeing this with our clients: administrative roles are evolving into agent supervision, strategic analysis, and continuous process improvement. The key is anticipation. Organizations investing now in AI, automation, and process design training are gaining a critical advantage.

From Tasks to Capabilities: A New Organizational Mindset

One of the most common mistakes we see is viewing agents as a one-to-one replacement for human jobs. That’s a reductive view. The real revolution lies in reimagining companies not as hierarchies of tasks, but as dynamic ecosystems of capabilities where humans and agents work as complementary peers.

The AI Agents and Agentic Workflows taxonomy (IALAB UBA, 2025) defines five levels of agent autonomy — from those requiring constant human intervention to those making complex decisions independently. This framework is vital for companies that want to scale AI responsibly and safely.

We use this model to build hybrid architectures where humans serve as strategic supervisors, agents handle execution and monitoring, and both interact through traceable and well-regulated workflows.

New Skills, New Leadership

This is not just a tech shift — it’s a cultural one. The most in-demand skills for 2030 won’t be repetitive tasks or narrow technical knowledge, but things like analytical thinking, tech literacy, continuous learning, adaptive leadership, and hybrid collaboration with AI systems.

The WEF predicts that 39% of today’s skills will be obsolete in five years, and 59% of workers will need some form of training. At Varegos, we support our clients not only with technology, but with adoption programs, training, and cultural transformation.

Risks and Dilemmas: Who Governs the Agents?

Like any major innovation, AI agents carry risks: decision-making bias, loss of human oversight, overreliance on automation, and lack of ethical standards. Multi-agent systems require solid governance, audit policies, and responsible design principles.

We advocate for open, traceable architectures with supervision rules and feedback loops, and we encourage including legal and ethical professionals in automation projects. It’s not enough for a system to work — it must be fair, safe, and explainable.

A Historic Opportunity

We’re at an inflection point. Just as electricity revolutionized industry in the 19th century, and the internet reshaped commerce in the 20th, AI agents are redefining how organizations operate, decide, learn, and scale.

At Varegos, we believe this isn’t about choosing between humans or machines — it’s about designing operating models where both coexist and enhance one another. Smarter companies, more human workers, faster decisions, and more flexible processes — that’s the future we’re helping build.

But that future won’t happen by default. It requires vision, strategy, investment — and above all, a genuine willingness to transform. That’s where we stand: as partners committed to guiding our clients through this next frontier.

The post Beyond the Hype: The True Impact of AI Agents on Business and Employment appeared first on Enterprise Viewpoint.