Campfire Audio and Pisen Launch New Audio and Charging Tech

Campfire Audio, based in Portland, Oregon, is a company worth knowing in the field of audio technology for the accuracy of its products. Its designers have succeeded in creating more than 70 different models of professional earphones, along with cables and accessories.

Wireless Earphones

The latest products tested were the wired earphones from the Iris and Axion models. Once the melodies flow into your ears, you will realize the quality the earphones offer.

• The Iris model, launched in late 2025, is an economical hybrid version of professional earphones for daily listening. If you are looking to enter the world of high-resolution audio, or as the company says, 'take your first steps in this world,' Iris was designed specifically for you. It features a durable and elegant design, with a transparent acrylic body printed using 3D technology, complemented by stainless steel artistic touches.

The Iris earphones, priced at $349, offer audio performance that fully meets expectations, delivering crystal-clear sound in both ears thanks to premium 10mm dynamic drivers. The earphones come with high-quality metal cables, a foldable storage case, three sets of ear tips (made of high-purity foam and silicone), along with a cleaning tool and cloth.

> The Axion model, another strong option from Campfire Audio, equipped with a silicon dynamic driver, is the first of its kind. This driver is designed to allow the Axion's compact housing to deliver a rich and impactful sound signature comparable to larger dynamic drivers.

Axion is characterized by being more suitable for on-the-go use, with equally clear sound and a distinct bass signature at both high and low volume levels. If forced to choose between it and the Iris model, I would choose the latter. However, if I hadn't experienced the Iris, Axion would have been the ideal choice with its smooth, natural sound that meets all my desires for my music playlists.

The Axion housing (priced at $249) is as eye-catching as the Iris, made of internally dyed transparent acrylic with touches of gold-plated stainless steel. The company designed them to be worn like regular earphones, or inverted with the wire behind the ear, as is common with professional monitoring earphones.

The earphone includes a dedicated high-quality USB-C cable that supports 'plug-and-play' functionality, featuring a built-in microphone and controls for hands-free calls and music playback. The cable also provides enhanced audio accuracy and more power to the earphones.

Included accessories include a zippered mesh carrying case, a dual-pocket mesh pouch for each earphone separately, three foam ear tips, three silicone ear tips, along with a cleaning tool and cloth.

Despite the differences between the two models, both offer an evolution from the first model and a solid build with all the necessary accessories. They come with a two-year manufacturer's warranty.

Website:

https://www.campfireaudio.com

Giant Charging Hub

> The 'Mega Charging Hub' from Pisen, with a capacity of 140W, offers a comprehensive solution for all charging needs on the go. It practically allows you to connect any device that needs power thanks to the variety of charging ports it is equipped with.

This desktop hub supports charging up to 8 devices simultaneously, including wireless charging for Apple Watches, earphones, iPhones, and other devices compatible with Qi2 and MagSafe technology. The ports include three USB-C ports, one USB-A port, in addition to two standard AC outlets, all in a single, direct-powered hub.

The hub can provide up to 140W through a single USB-C port. It features a built-in, gently pulsing flow lighting system to indicate battery charging status. The screen displays a three-color system for adaptive voltage measurement, showing green at 5V, yellow between 9-15V, and purple at 20V.

Safe charging is provided by GaN (Gallium Nitride) technology, with 9 built-in protection systems to safeguard charged devices from various risks, most notably over-temperature and short-circuiting. The top magnetic wireless charging panel also offers a rotation angle of up to 65 degrees, allowing for viewing the smartphone in both horizontal and vertical orientations while charging. This Mega Charging Hub is available in black or yellow, and comes with a 6-foot, 240W USB-C to USB-C fast charging cable.

Website:

https://www.pisen.us/products/pisen-all-in-one-charger

• Tribune Media Services

At the Dell Technologies World conference, which concluded Thursday in Las Vegas, the most pressing question about artificial intelligence is no longer whether organizations have started their experiments, but why many are struggling before they translate into operational value within their businesses. The rapid experimentation phase is no longer sufficient to prove the technology's viability, especially as companies move from simple smart assistants to agentic AI systems capable of executing multiple steps, interacting with internal data and systems, and potentially impacting entire workflows.

In a special interview with Asharq Al-Awsat, Scott Belsky, Vice President of Professional Services at Dell Technologies, places this transformation within a practical framework. Organizations, as he explains, often start with a 'proof of concept' to verify the validity of a specific solution or use case, but then require a different phase he calls 'proof of value.'

This phase goes beyond asking: Does the technology work? It delves into deeper questions: What value will it generate? What are the performance indicators? What results does the organization want to achieve? Belsky says this phase typically lasts from four to six weeks, and involves using the organization's own data, defining success metrics, and then testing the use case before moving to full-scale deployment.

Proof of Value

This distinction between 'proof of concept' and 'proof of value' encapsulates one of the most important lessons in enterprise AI today. The success of a limited technology experiment does not necessarily mean it is ready for daily workflow integration. Belsky clarifies that what makes an agentic AI experiment transferable to a real workflow is defining that workflow itself, and then identifying the performance indicators and desired outcomes. If 'proof of value' does not demonstrate that the opportunity is worth scaling, it may indicate that the organization does not need to proceed to production in this case, or that it should seek another, more viable use case.

From here, the path to production becomes less about purchasing infrastructure alone and more about the hidden work that happens afterward. Belsky notes that organizations often underestimate the amount of work required after implementing a solution, including maintaining the modernity of the entire system, from the control and orchestration layer to the applications and use cases built on top of it. However, he sees one of the biggest integration challenges emerging around how an organization identifies suitable datasets, prepares and cleans them, builds data pipelines, and secures them from end to end.

Redesigning Workflows

This point makes enterprise AI as much a management and operational project as it is a technical one. An organization that buys AI infrastructure without defining where its data resides, its quality, and how it will enter the workflow, will often face a gap between theoretical capability and actual operation. That is why Scott Belsky emphasizes that choosing a use case should not start with a small, isolated task, but with understanding an entire business process or a specific function from beginning to end. He warns that focusing on overly tactical use cases may prevent the organization from capturing the greater benefit, which is redesigning the entire process, especially in the context of agentic AI.

What separates a productive use case from a technical experiment, according to Belsky, is that it becomes 'actually used in operations' by business users, and ultimately produces results and value. In the proof of value stage, testing might be on a subset of data or a limited group of users. True production, however, means scaling the use to a broader scope within the organization, after addressing aspects of implementation, integration, operation, security, and governance.

Complexity of Agents

These requirements become more complex when an organization moves from a smart assistant that answers a question to an agent or multiple agents interacting with the workflow. A traditional assistant is often closer to a single use case involving a request, an answer, and a specific outcome. Agentic AI, however, can require coordination between multiple agents, communication between them, and linkage to different workflows. Here, the question is not just whether the model can accomplish the task, but whether the organization can coordinate these agents, secure them, govern them, and control their permissions? Belsky states that this can become 'very complex' when it comes to one agent communicating with another, securing that communication, and ensuring appropriate access controls across a multi-agent workflow.

Usable Data

Data once again emerges as the center of difficulty, with the recurring question within organizations being: Is our data ready for AI? Belsky answers that it is difficult to judge generally. He believes that one cannot look at the entire data footprint within an organization, including structured and unstructured data, and simply say it is ready or not ready. Readiness must be measured in the context of the targeted use case or agentic outcome: Can this data support the specific use? And can it be integrated and presented in a way that achieves the desired result?

This is where professional services come in. Once the organization defines the use case, services can help identify suitable datasets, assess their quality, determine if they need enrichment, refinement, or improvement, then prepare data pipelines, integrate them into the use case, and govern and manage them over time from a compliance perspective. This explains why moving to production is not just about deploying a model or buying servers, but a series of decisions about data, processes, integration, and security.

Data Bottleneck

This aligns with one of the main themes of Dell's announcements at the conference. The company focused on expanding the 'Dell AI Factory with NVIDIA' and the 'Dell AI Data Platform' to address the problem of transforming scattered data within organizations into usable data for AI applications. Updates include indexing billions of unstructured files, linking them to governed data pipelines, accelerating SQL analytics by up to 6 times using GPUs, and faster vector indexing by up to 12 times. These figures do not serve an abstract technical presentation but are directly related to what Belsky describes: that a data bottleneck can quickly turn into a bottleneck in transitioning from experimentation to production.

Choosing the Right Model

As the ecosystem expands around open and closed models and various partners, the issue of choosing the right model becomes part of reducing deployment risks. Dell announced at the conference, which Asharq Al-Awsat attended, the expansion of its ecosystem through partnerships including model and platform providers such as Google, Hugging Face, OpenAI, Palantir, ServiceNow, and SpaceXAI. However, the practical value of these partnerships becomes apparent when linked to the actual use case. Belsky points out that Dell helps clients identify the most suitable model or large language model, then consider performance, secure the environment, protect data and the model, and optimize performance after deployment.

Scott Belsky does not pit data and computing against each other. When asked which has become more important, data orchestration or computing, he said the choice is difficult because 'computing without data yields no result,' and likewise, 'data without computing yields no result.' However, he adds that an organization may possess 'the best computing in the world,' yet it will not achieve the desired results if the quality of the underlying data is poor.

AI Sprawl

Cost emerges as another part of the implementation gap. As experiments expand across different departments, a state of 'AI sprawl' may arise within the organization, where multiple teams use different tools and models without a clear view of consumption volume. Belsky notes that many clients do not actually know their current token consumption or the associated costs. Therefore, before comparing on-premises or cloud operations, the first step is to understand the current cost and consumption footprint, then measure it against alternative operating models such as 'Deskside Agentic AI' or within the data center.

In this context, some of Dell's announcements regarding 'Deskside Agentic AI' are linked to the idea of controlling costs, not just running AI on a workstation. The company states that some continuous agentic workloads may become more economically predictable when run locally or within an organization-controlled environment, rather than relying entirely on consumption-based cloud APIs.

This leads to the point of risk in automating poorly designed workflows. Agentic AI may accelerate what exists, but if it enters an unclear or fragmented process, it may accelerate complexity itself. Belsky stresses that the start must be with understanding the current workflow, then defining its future desired form, and thinking about it flexibly, because the agents themselves will evolve, and workflows will change as technology advances. This means the design should not be rigid but adaptable as agent capabilities and use cases change.

Sovereignty and Sensitive Sectors

In the context of the Middle East and Saudi Arabia, Belsky does not see the implementation pattern differing radically from Europe or North America in terms of basic steps like identifying use cases, implementing an 'AI factory,' and driving clients toward measurable results. However, he acknowledges that data sovereignty issues are 'clearly important,' and that working in the region, including Saudi Arabia, involves managed services, security services, and operating models that help clients address these requirements. He says the general pattern of helping clients achieve results by starting with the right use cases appears similar in Saudi Arabia and the Middle East compared to other markets.

However, the complexity of data, sovereignty, and security makes some sectors more sensitive when transitioning to production. Scott Belsky links the most challenging sectors to data protection, security, and sovereignty challenges. Entities dealing with sensitive or regulated data, such as government, finance, health, and energy, may find themselves compelled to build more controlled capabilities, and perhaps rely more on internal or managed environments, because data protection becomes part of the deployment decision, not a later detail. He adds that sectors where data sovereignty, protection, and security are the biggest challenge also represent a significant opportunity, as they drive these organizations toward more controlled operating environments.

Skills Challenge

The AI skills gap remains a constraint in regions like Central and Eastern Europe, the Middle East, and Africa, while Saudi Arabia sets ambitious goals for human capital development. In this context, Saudi investment in workforce development, from the programs of the Saudi Data and Artificial Intelligence Authority (SDAIA) to universities and their partnerships with the private sector, stands out as a clear framework within which international partners can operate.

Information provided by Dell indicates local initiatives, including its collaboration with SDAIA to train cloud computing specialists, as part of digital enablement efforts for government entities and achieving the goals of 'Saudi Vision 2030.' This also includes the 'NextGen Sales Academy' program to qualify new professionals through a two-year track, the 'Academic Alliance' program in collaboration with over 15 universities to provide technical training within curricula, and the 'Global Services Associate' program for training and employing graduates. Dell, Aramco, and the National Information Technology Academy also signed an agreement in 2024 to empower local talent with advanced skills in science and technology and develop a deployable technical talent base in the Kingdom.

These initiatives do not seem marginal in the context of enterprise AI. If organizations are required to move from experimentation to production, the skills gap becomes part of the implementation gap itself. Success requires teams that understand the domain, data, governance, security, and workflows, not just the ability to use a ready-made AI model. Agentic AI, in particular, increases the importance of these skills because it adds not just a new tool, but changes the way work is done.

However, Belsky does not present the path to production as a fixed or final route. He sees the challenges as clear, including workflows, data, security, and governance. The answers themselves will evolve with the evolution of agents and technology, meaning organizations need an updatable operating model, not a rigid one-time plan.

Execution Discipline

Belsky believes that the organizations that will succeed are those that rethink workflows and processes in the context of agentic AI. He describes AI as similar to previous transformations from at least one angle, saying, 'It's all about transforming processes and changing the way business is done.' He adds that organizations that do not become AI-centric and do not rethink all their workflows within this context 'will face severe challenges.' Success does not come from isolated experimentation, but from the ability to design processes around