Category Robots Podcast

When 3-year-old Cameron arrived at the UC Davis MIND Institute for a recent medical appointment, she was in for a surprise: cuddly, social AI robots visiting from Japan were in the waiting room. "We both melted when we saw them," said Cameron's mom, Caroline Nguyen. "They're so adorable, and the little sounds they make—I mean, she took to them right away."

Autonomous drones could deliver packages, inspect bridges and skyscrapers, monitor emergencies like wildfires and, eventually, ferry people. But the airspace still lacks the testing and coordination infrastructure needed to prove drones can operate safely around people, buildings, aircraft and each other.

Imitation learning is changing how industrial robots are trained, shifting from rigid programming to learning through real-world interaction. Anders Billesø Beck explains why data quality, force and production-grade hardware matter.

DataRobot now supports the Agentic Resource Discovery Specification, making DataRobot Agent Skills and MCPs easier for AI clients, registries, and developers to find.

Agents are only as useful as the capabilities they can reach.

A coding agent can write code. A workflow agent can call tools. An enterprise agent can reason across systems. But all of that depends on the same basic question: when the agent needs a capability, how does it find the right one?

Until now, the answer has mostly been manual. Developers wire in MCP servers, install skills, point agents at docs, and maintain long lists of tools that may or may not be relevant to the task at hand. That works for a small number of hand-picked integrations. It breaks down when every platform, team, and community is publishing new agentic resources.

That is why we are excited to share that DataRobot now supports the Agentic Resource Discovery Specification, also known as ARD.

DataRobot now publishes an ARD-compatible AI catalog for DataRobot Agent Skills and MCP Servers, making those skills and MCPs discoverable from our domain through the standard .well-known/ai-catalog.json path at https://datarobot.com/.well-known/ai-catalog.json

Why ARD matters

Agentic Resource Discovery is an open specification for publishing, discovering, and verifying agentic resources across the web. Those resources can include skills, MCP servers, APIs, agents, tools, workflows, and other capabilities.

The model is simple: providers publish a catalog of available resources under their own domain. Discovery services and AI clients can then find, index, and resolve those resources when an agent needs them.

That matters because the agent ecosystem is moving from static wiring to dynamic discovery.

Instead of asking developers to preload every possible tool and skill into an agent’s context, ARD gives agents and registries a standard way to discover the right capability for the task. The agent can search, select, and connect to relevant resources without carrying every integration by default.

For enterprises, that discovery layer is especially important. Teams need agents that can find useful capabilities, but they also need control over what gets surfaced, where it comes from, and how it is governed.

What DataRobot is publishing

DataRobot’s ARD catalog currently points to DataRobot Agent Skills and MCPs.

This includes skills for:

• Model training
• Model deployment
• Predictions and batch scoring
• Feature engineering
• Model monitoring
• Model explainability
• Data preparation
• App Framework CI/CD
• External agent monitoring
• Agent Assist

These skills package DataRobot platform knowledge into task-scoped context that coding agents can use directly. They help agents understand DataRobot workflows, SDK patterns, deployment steps, validation checks, and observability practices.

In other words, they teach agents how to use DataRobot correctly.

With ARD support, those skills are not only available in repositories and agent environments. They are also published in a standard catalog that discovery tools can crawl, index, and resolve.

From installable skills and MCPs to discoverable platform context

We have been investing in DataRobot Skills and MCPs because agents need more than documentation. They need operational context.

A human developer can read docs, infer missing steps, ask a teammate, and recover when an API call fails. An agent needs the right context at the right moment. Otherwise, it guesses.

Skills and MCPs reduce that guesswork by giving agents precise instructions for common platform workflows. ARD takes the next step by making those resources easier to find.

That shift matters for developer experience. It also matters for platform teams.

If you are building agents on DataRobot, you should not have to manually teach every tool where DataRobot skills and MCPs live. If you are building an AI client or registry, you should have a standard way to discover DataRobot resources. If you are governing agentic AI inside an enterprise, you should be able to decide which catalogs and registries your agents can use.

ARD gives the ecosystem a path toward that model.

Try it

• Learn more through the official ARD specification.
• Get started today: Follow the quickstart guide to publish your AI catalog in minutes
• Build: Create your first DataRobot Agent
• Read: Learn more about DataRobot Skills, MCP, and the agentic developer surface

What comes next

Agentic discovery is still early, and the specification is moving quickly. That is exactly why we wanted DataRobot to participate now.

The agentic web will not be built from one marketplace, one vendor catalog, or one hard-coded tool list. It will need open discovery, clear ownership, and resources that agents can actually use.

DataRobot’s role is to make enterprise AI agents easier to build, operate, monitor, and govern. Supporting ARD is another step toward that future: DataRobot platform context that is not just available, but discoverable.

Agents should not have to guess where the right capability lives.

Now, they can find DataRobot.

The post DataRobot Agent Skills and MCPs are now discoverable through Agentic Resource Discovery appeared first on DataRobot.

A new chip developed by MIT researchers could help tiny, low-power UAVs avoid obstacles as they zip around tight corners inside an industrial HVAC system to check for gas leaks. The chip allows small autonomous robots and other battery-limited devices to construct detailed 3D maps of their environments in real time using only about as much power as a single LED. A robot could use such a map to plan a collision-free path to reach its goal.

Image credit: By Jeff Dahl – Own work by uploader, Based on the public domain document: [1], CC BY-SA 3.0, Link

By Stephanie Parker

This research from the NeuroAI Lab of Martin Schrimpf, part of EPFL’s Schools of Computer and Communication Sciences and Life Sciences, uses AI models to predict exactly where to stimulate the brain to evoke images of faces and specific objects in the users instead of simply evoking spots of light. The models developed at EPFL were used by Dutch researchers for live trials on sighted monkeys. The preliminary results, presented in April at the International Conference on Learning Representations, show very promising implications for vision in humans as well.

“The motivation for this project is that there are many people with visual deficits that are irreparable, in the sense that somewhere along the visual processing stream, starting with the retina, there is a deficit which cannot be repaired,” says Johannes Mehrer, a scientist in the NeuroAI lab who led the research. “One way of tackling this problem is to develop a visual prosthesis.”

There are multiple kinds of visual prosthetics including retinal, optical nerve, and cortical. Retinal prosthetics are placed somewhere on the retina, and optical nerve prosthetics are used when the retina is too damaged for an implant and the optical nerve can be stimulated instead. When neither the retina nor optical nerve can be implanted, cortical prosthetics are used. These bypass the retina and optical nerve entirely and work instead by stimulating the visual cortex, using electrodes to “draw” images onto it. However, thus far, this approach is limited in that it targets lower-level regions of the brain where it is only possible to project light flashes and simple shapes. There are also hardware constraints because multiple electrodes are needed to stimulate different areas at the same time, but only a certain number of electrodes can be used in one area.

“The images they can elicit, in this case simple symbols, are really limited in their complexity,” Mehrer explains. “At the moment, existing approaches to visual prostheses couldn’t elicit the percept of a more complex visual object such as a house or a car.”

Higher-level visual regions of the brain underlie the processing of more complex objects and could thus serve as a target for a new generation of visual prostheses allowing for eliciting images of faces, houses, and other objects. However, these higher-level regions are less accessible, because it is not known exactly where and how to stimulate them. This is where the AI model comes in.

Towards restoring meaningful sight

“We had the idea to use an artificial neural network, in this case a specific type called a topographic neural network, to test various patterns of brain stimulations in these higher-level regions of the brain and simulate their outcomes,” Mehrer says. “We can then run all sorts of simulations using different combinations of different parameters that would otherwise take up a lot of experimental time and would cost a lot of money.”

The EPFL researchers, working entirely on computers, set up a model to select the best combination of images with the specific pattern of stimulation. Following their results, a team of resarchers in Amsterdam decided to test the model’s prediction on two of their monkeys who already had implants for other experiments not involving EPFL.

“Our model turned out to be quite efficient in predicting which stimulation pattern would yield a strong effect on the monkeys’ behavior with respect to visual object recognition,” says Martin Schrimpf, head of the NeuroAI Lab. “Our models can do the image selection, but the more crucial part is that given an image, it can tell us what the optimal stimulation pattern for a particular desired behavior is.”

What the researchers have been able to show so far with this work is that they can shape object perception, meaning that if a visual stimulus is presented, they can bias its representation in the brain. However, they cannot yet create object perception out of nothing. Stimulating the cortex while there is no visual stimulus presented would be their next step towards restoring meaningful sight to the blind.

“The monkey saw an image already, and then we were able to basically distort it to change the perception in somewhat predictable ways,” says Schrimpf. “The bigger goal will be to evoke a percept from scratch: to make someone see something meaningful even when their eyes aren’t delivering a usable image.”

This work showing that model-guided brain stimulation could lead to more advanced visual prosthetics could also be applied to hearing prosthetics. Through a grant from the Horton Health Foundation, Schrimpf and his team will next investigate if this kind of modeling works for auditory stimulation.

“Cochlear implants are great, but they are also not perfect in many ways, and they don’t really fully restore auditory processing,” says Schrimpf. “Our idea is to develop these kinds of topographic models that can predict what stimulation does to neural activity for auditory processing as well.”

The Future of Manufacturing: How AI Is Driving Efficiency and Innovation?

Manufacturing is entering a new phase of digital transformation where Artificial Intelligence (AI) is becoming part of everyday operations. Global manufacturers are facing increasing pressure to improve productivity, reduce costs, manage supply chain disruptions, and maintain high product quality. Traditional approaches alone are no longer enough to keep pace with changing market demands.

According to Deloitte’s Smart Manufacturing Survey, manufacturers continue to increase investments in AI and smart factory technologies to improve predictive maintenance, product quality, and supply chain resilience.

AI is helping manufacturers address these challenges by turning operational data into actionable insights. AI solutions for manufacturing support predictive maintenance, quality inspection, intelligent document processing, and supply chain optimization across the manufacturing ecosystem.

Why Is AI Important in Manufacturing?

Organizations looking to accelerate digital transformation are increasingly investing in AI-powered manufacturing solutions that improve visibility across operations and support data-driven decision-making.

Modern manufacturing environments generate large volumes of data from machines, sensors, enterprise applications, maintenance records, supplier networks, and operational documents. Much of this information remains underutilized because it exists across disconnected systems.

AI helps organizations connect these data sources, identify patterns, and support faster decision-making. As a result, manufacturers can reduce downtime, improve resource utilization, and respond more effectively to operational changes.

Organizations that invest in AI are also building more resilient operations that can adapt to market fluctuations and customer expectations.

What Are the Benefits of AI in Manufacturing?

AI helps manufacturers improve productivity, reduce downtime, optimize supply chains, and automate business processes while enabling faster, data-driven decision-making.

Industry adoption continues to accelerate. According to Deloitte’s 2025 Smart Manufacturing Survey, manufacturers reported up to 20% improvement in production output, 20% improvement in employee productivity, and 15% unlocked operational capacity through smart manufacturing initiatives.

The same study found that 80% of manufacturing executives plan to invest at least 20% of their improvement budgets in smart manufacturing technologies over the next few years.

Manufacturers are using AI to:

• Reduce equipment downtime
• Improve product quality
• Optimize inventory management
• Strengthen supply chain visibility
• Increase workforce productivity
• Automate document-intensive processes
• Lower operational costs
• Support data-driven decisions

As AI adoption grows, manufacturers are moving beyond isolated pilot projects and integrating intelligent technologies across production, maintenance, supply chain, and enterprise operations. Organizations that build strong data and automation foundations today will be better positioned to compete in the future of smart manufacturing.

Top AI Use Cases in Manufacturing

1.     Predictive Maintenance

Manufacturers adopting AI use cases for predictive maintenance and equipment management can reduce unplanned downtime and improve asset reliability.

Unexpected equipment failures can disrupt production schedules and increase operational costs. AI models analyze machine performance data to detect early signs of wear and identify potential failures before they occur.

This allows maintenance teams to schedule repairs proactively, reduce unplanned downtime, and extend the life of critical assets.

2.     Intelligent Quality Control

Advanced computer vision solutions for manufacturing help organizations automate defect detection and strengthen quality assurance processes.

Manual quality inspections can be time-consuming and inconsistent, particularly in high-volume production environments.

AI-powered computer vision systems can analyze products in real time, identify defects, and maintain quality standards across production lines. Faster defect detection helps reduce waste and minimize costly rework.

3.     Supply Chain Optimization

Supply chain disruptions continue to challenge manufacturers across industries. AI-powered supply chain management can analyze demand patterns, supplier performance, inventory levels, and logistics data to support more accurate forecasting.

Better visibility across the supply chain helps organizations improve inventory management, reduce delays, and maintain business continuity. Many manufacturers are also adopting AI-powered supply chain solutions to improve forecasting and operational coordination.

4.     Production Planning and Workforce Management

AI can evaluate multiple production variables simultaneously, including workforce availability, machine capacity, inventory levels, and customer demand.

This enables manufacturers to optimize production schedules, improve workforce allocation, and reduce operational bottlenecks.

5.     Inventory and Facility Management

Manufacturers often struggle with excess inventory, stock shortages, and facility management challenges. AI can help organizations optimize inventory levels, monitor asset utilization, and improve operational planning across manufacturing facilities.

How Does AI Improve Manufacturing Documentation?

Operational efficiency depends on accurate and accessible information. Manufacturing organizations manage thousands of documents, including work orders, maintenance logs, inspection reports, compliance records, engineering drawings, supplier contracts, invoices, and standard operating procedures.

Managing these documents manually can slow down workflows and create information gaps.

At USM, we help manufacturers modernize document-intensive operations through AI-powered automation capabilities that include:

• Intelligent document processing
• Automated data extraction
• AI-assisted document classification
• Enterprise search and knowledge retrieval
• Workflow automation for operational documentation
• Integration with existing ERP and enterprise platforms

By reducing manual effort and improving access to information, organizations can accelerate decision-making and improve operational consistency.

How USM Supports AI-Driven Manufacturing?

USM works with manufacturing organizations to transform data-intensive and document-heavy business processes through practical AI solutions. Our expertise spans AI in Manufacturing, intelligent automation, predictive analytics, and connected factory initiatives.

Our manufacturing AI capabilities support use cases such as:

• Predictive equipment maintenance
• Intelligent supply chain management
• AI-powered inventory optimization
• Workforce management automation
• Facility management solutions
• Customer and operational analytics
• Enterprise knowledge management
• Agentic AI solutions for manufacturing operations

These capabilities help organizations reduce operational inefficiencies while improving visibility across business functions.

Conclusion: Building the Future of Manufacturing with AI

The future of manufacturing will be shaped by organizations that can combine operational expertise with intelligent technology.

As manufacturers continue to modernize their operations, AI will play an increasingly important role in improving productivity, reducing operational complexity, and enabling smarter business decisions.

At USM – best AI company in USA, we help manufacturers modernize document-heavy workflows, automate operational processes, and build AI-powered manufacturing ecosystems that improve visibility and reduce manual efforts. Our AI manufacturing solutions are designed to help organizations build more resilient, efficient, and future-ready operations.

 

Contact us to know more about How AI Is Driving Efficiency and Innovation? Book Executive AI Briefing →

 

Frequently Asked Questions

• What Is AI in Manufacturing?

AI in manufacturing is the use of artificial intelligence technologies to automate processes, analyze operational data, improve production efficiency, predict equipment failures, and support better business decisions across the manufacturing lifecycle.

• How does AI improve manufacturing efficiency?

AI improves efficiency by automating repetitive processes, predicting equipment failures, optimizing production schedules, improving inventory management, and helping organizations make faster decisions using operational data.

• Can AI reduce manufacturing costs?

Yes. AI can help reduce costs by minimizing downtime, improving quality control, lowering maintenance expenses, reducing waste, and streamlining document-intensive workflows.

• Is AI only for large manufacturers?

No. AI solutions are increasingly scalable and can be implemented across organizations of different sizes. Many manufacturers begin with targeted use cases and expand adoption as they realize business value.

• What are the top AI use cases in manufacturing?

The most common AI use cases include predictive maintenance, quality inspection, supply chain optimization, inventory management, production planning, document automation, and workforce management.

• Can AI integrate with ERP systems?

Yes. Modern AI platforms can integrate with ERP, MES, CRM, and other enterprise systems to automate workflows and improve operational visibility.

• How does AI support smart factories?

AI supports smart factories by connecting machines, sensors, enterprise systems, and operational documents to provide real-time insights, improve productivity, and enable data-driven decisions.

• How does USM help manufacturers adopt AI?

USM helps manufacturers implement AI solutions for predictive maintenance, document automation, intelligent supply chain management, enterprise knowledge management, and workflow optimization. Our AI capabilities integrate with existing enterprise systems to improve operational efficiency and support digital transformation initiatives.

In this article, Patrick Schick, Product Manager for 3D Vision at IDS, explains the background and technical decisions behind Nion, puts iToF technology into context, and highlights typical areas of application.

Teams are moving AI agents from prototype to workflow fast. One agent gets connected to a document store. Another starts calling internal tools. A third begins touching customer data. 

Soon, agents are operating across systems before governance teams have a clear record of what they can access, who owns them, or what they’ve done.

AI agents can retrieve information, call tools, trigger workflows, and act across business systems. When they operate outside approved governance workflows, they create an ungoverned operational layer inside the enterprise that can expose sensitive data, bypass policy controls, and make incident response harder.

To find and govern unsanctioned AI agents, enterprises need to:

• Identify where agent activity already exists
• Determine what each agent can access
• Assign clear ownership and scope
• Apply runtime monitoring, audit trails, and policy controls

The goal isn’t to shut down experimentation. It’s to make the governed path easier than the workaround. That starts with visibility: knowing which agents exist, what they can do, which systems they touch, and whether their actions can be reviewed after the fact.

Key takeaways

• Shadow agents are unsanctioned AI agents that operate outside approved governance, security, or deployment workflows.
• They often emerge when teams can prototype agents faster than the enterprise can govern them.
• The biggest risk is unmonitored action across tools, data, APIs, and workflows.
• Enterprises need a reliable inventory of which agents exist, who owns them, what they can access, and what actions they can take.
• Effective governance brings agents under identity, scope, permissions, monitoring, and auditability.
• The governed path should be clear enough and practical enough that teams do not need workarounds.

What are shadow agents in enterprise AI?

Shadow agents are AI agents that operate outside an enterprise’s approved governance, security, or deployment workflows. They often begin as prototypes, internal automations, or team-level tools, then expand into production workflows without a central inventory, assigned owner, defined permission model, or audit trail.

The risk increases when a shadow agent connects to enterprise systems. That can include document repositories, customer databases, ticketing systems, internal APIs, model context protocol (MCP) servers, workflow tools, or other agents. 

Once an agent can access data, call tools, or trigger actions, it needs the same governance attention as any other system operating on behalf of the business.

Shadow agents can include:

• A developer-built agent that calls internal APIs without formal approval
• A workflow agent connected to customer data before security review
• An internal assistant that retrieves sensitive documents without access controls
• A team-level automation that uses shared credentials or undocumented permissions
• An agent prototype that quietly becomes part of a live business process

The central issue is visibility. Enterprises can’t govern agents they can’t see. Before teams can evaluate risk, enforce policy, or investigate behavior, they need a reliable record of which agents exist, what they’re connected to, what permissions they have, and what actions they’ve taken.

Why do shadow agents appear in enterprise AI environments?

Shadow agents appear when teams can build and connect AI agents faster than the enterprise can govern them. Prototyping is easy, business teams are under pressure to show AI value, and governance processes often feel slower than the work teams are trying to get done.

Most shadow agents don’t start as a deliberate attempt to bypass controls. They usually start as practical experiments: a developer testing an agent, a team automating a workflow, or a business unit connecting an assistant to internal data. The risk grows when those experiments keep expanding without a formal path into governed deployment.

Cause	How it creates shadow agent risk	How to respond
Fast prototyping	Teams connect agents to tools, data, or workflows before production governance is defined.	Require agent identity, scope, and access review before agents connect to live systems.
Pressure to prove AI value	Teams prioritize speed and visible outcomes over access controls, monitoring, and documentation.	Create a faster approved path for governed agent deployment.
Late governance review	Security and governance teams discover agents after they’re already connected to enterprise systems.	Embed governance checks into design, testing, and deployment workflows.
No central inventory	The enterprise can’t see which agents exist, who owns them, or what they can access.	Maintain a centralized inventory of agents, owners, tools, data sources, and permissions.
Unclear deployment standards	Teams don’t know when an experiment has crossed into production use.	Define clear thresholds for when agent prototypes require formal governance review.
Friction in approved workflows	Teams create workarounds when the governed path feels slower than the unofficial path.	Make compliant deployment easier to follow, monitor, and repeat.

Shadow agents are often a process problem before they’re a technology problem. When teams don’t have a clear, fast, and practical way to deploy governed agents, they create their own path. Effective agent governance closes that gap by making approved deployment easier to follow, easier to monitor, and easier to scale.

Why are shadow agents risky?

Shadow agents are risky because they can act inside enterprise systems without the visibility, permissions, monitoring, and audit trails required to control that behavior. An unsanctioned AI agent may access sensitive data, call internal tools, trigger workflows, or pass information to another system before governance teams know it exists.

That makes shadow agents different from ordinary software sprawl. A forgotten app may create security exposure. A shadow agent can create security exposure and take action. It can interpret a request, retrieve context, choose a tool, and execute a step inside a workflow. If that behavior is not governed, the enterprise may not know what happened, why it happened, or how to prevent it from happening again.

Shadow agents can access sensitive data

Many agents become useful because they connect to enterprise data. That same connection creates risk when access is not scoped, approved, or monitored. A shadow agent may retrieve customer records, employee data, financial information, proprietary documents, or regulated data without the right controls in place.

Shadow agents can take action across systems

AI agents can do more than return answers. They can call APIs, update records, create tickets, send information to other tools, or trigger downstream workflows. When those actions happen outside approved governance workflows, small errors can become business problems quickly.

Shadow agents can be hard to investigate

When an incident happens, teams need to reconstruct what the agent did. That requires logs of inputs, outputs, retrieved context, tool calls, actions, and outcomes. Without that audit trail, security, compliance, and operations teams are left piecing together behavior after the fact.

The core risk is traceability. Enterprises need to know which agents exist, what they can access, what actions they can take, and whether their behavior can be reviewed. Without that record, shadow agents create blind spots across security, compliance, and operations.

How can enterprises find shadow agents?

Enterprises can find shadow agents by looking for agent behavior across tools, data sources, APIs, and workflows. Many shadow agents won’t appear in a central AI inventory because they started as experiments, scripts, assistants, or team-level automations.

Governance, security, IT, and AI teams should start by reviewing the environments where agents can connect to live business systems. That includes developer workspaces, cloud environments, automation platforms, internal applications, copilots, model context protocol (MCP) servers, and business-unit workflows.

Useful discovery questions include:

• Which AI agents or LLM applications are connected to enterprise data?
• Which agents can call internal tools, APIs, or workflow systems?
• Which agents use shared credentials, service accounts, or unmanaged permissions?
• Which prototypes are now part of recurring business processes?
• Which agents have no assigned business owner or technical owner?
• Which agents lack logs for inputs, outputs, tool calls, actions, and outcomes?

The goal is to create a working inventory that shows which agents exist, who owns them, what systems they touch, what permissions they have, what actions they can take, and whether their behavior can be reviewed after the fact.

How can enterprises govern shadow agents once they find them?

Enterprises can govern shadow agents by bringing them into a formal agent governance workflow. That process should clarify what the agent does, who owns it, what systems it can access, what actions it can take, and how its behavior will be monitored over time.

The first step is classification. Some shadow agents may be useful and worth governing. Others may be too risky, redundant, or poorly designed to keep in place. Governance teams should evaluate each agent based on business value, system access, data sensitivity, autonomy level, and auditability.

How do you assign ownership for an AI agent?

Every agent needs a business owner and a technical owner. The business owner is accountable for the use case, expected outcome, and acceptable risk. The technical owner is accountable for implementation, access, monitoring, and maintenance.

Ownership matters because agents can act across workflows. If an agent behaves unexpectedly, the organization needs to know who can review it, restrict it, update it, or shut it down.

How do you define what an AI agent can access and do?

A shadow agent should not keep whatever access it gained during experimentation. Governance teams need to define the agent’s purpose, approved systems, allowed actions, and off-limits data.

The permission model should match the job the agent is supposed to perform. An agent that summarizes support tickets does not need the same access as an agent that updates customer records or triggers account changes.

How do you monitor and audit AI agent behavior?

Governance teams need a record of agent behavior in production. That includes inputs, outputs, retrieved context, tool calls, actions, and outcomes. These records help teams investigate incidents, validate policy compliance, and understand how agent behavior changes over time.

A governed agent should be reviewable. Teams should be able to reconstruct what happened, which tools were used, what data was accessed, and which action the agent took.

How do you decide whether to govern, restrict, rebuild, or retire a shadow agent?

Once a shadow agent is evaluated, teams can choose the right response. A useful agent with manageable risk may be moved into an approved governance workflow. A high-risk agent may need tighter permissions, additional monitoring, or a redesigned workflow. An agent with unclear ownership, weak controls, or low business value may need to be retired.

The standard should be simple: if an agent can access enterprise systems or act on behalf of the business, it needs identity, ownership, scoped permissions, monitoring, and auditability.

Learn how to govern agentic AI across the full lifecycle

Shadow agents are one warning sign of a larger governance challenge. As enterprises move from isolated AI experiments to agentic systems that retrieve information, call tools, trigger workflows, and act across business systems, governance has to become part of how agents are built and operated.

The enterprise guide to agentic AI governance explains how to govern AI agents across the full lifecycle, including permissions, audit trails, runtime monitoring, lifecycle controls, and fleet-level oversight.

Read the ebook to learn how to build the governance foundation for agentic AI at enterprise scale.

FAQ

What are shadow agents in enterprise AI?

Shadow agents are AI agents that operate outside approved governance, security, or deployment workflows. They may access data, call tools, trigger workflows, or support business processes without a central inventory, assigned owner, defined permission model, or audit trail.

Why do shadow agents appear?

Shadow agents appear when teams can build and connect agents faster than the enterprise can govern them. They often begin as prototypes, automations, or team-level tools, then expand into real workflows before security, compliance, or governance teams have full visibility.

Why are shadow agents risky?

Shadow agents are risky because they can access sensitive data, call internal tools, and take action across enterprise systems without approved controls. If they lack monitoring and audit trails, teams may not be able to reconstruct what happened after an incident.

How can enterprises find shadow agents?

Enterprises can find shadow agents by looking for agent behavior across tools, data sources, APIs, automation platforms, cloud environments, MCP servers, and business workflows. The goal is to identify which agents exist, what they connect to, who owns them, and whether their behavior can be reviewed.

How should enterprises govern shadow agents?

Enterprises should govern shadow agents by assigning ownership, defining scope, reviewing permissions, adding runtime monitoring, and capturing audit trails. Each agent should have a clear purpose, approved access, documented controls, and a reliable record of its actions.

The post Shadow agents: find and govern unsanctioned AI agents appeared first on DataRobot.

Current flying objects face a trade-off: Drones with propellers, for instance, are very agile and able to hover; however, they use up a lot of energy. Airplanes, on the other hand, feature fixed wings that allow them to fly very efficiently. The downside: They can't remain suspended in the air like a kestrel on the lookout for prey.

Successfully integrating robotics with HF welding machines requires more than installing equipment. Robots can only accelerate manufacturing cycles, protect workers from hazards and eliminate errors like burn-through when implemented properly.

What Is Physical AI? How AI Moves From Chatbots to Robots Artificial intelligence has made remarkable progress in recent years. Large language models can answer questions, generate code, write articles, and engage in complex conversations. However, most AI systems still exist entirely in the digital world, processing information, generating outputs, and interacting solely through screens […]

One of only four teams remaining from more than 130 competitors worldwide, our team AURA Foresight is developing autonomous technology to stop wildfires before they grow out of control.

AURA Foresight has been selected as a finalist in the prestigious XPRIZE Wildfire Autonomous Wildfire Response competition, emerging as one of just four teams remaining from more than 130 teams from around the world.

XPRIZE Wildfire is a four-year, US$11 million global competition designed to accelerate breakthrough technologies capable of ending destructive wildfires. The Autonomous Wildfire Response track, worth US$5 million, challenges teams to autonomously detect, verify and respond to wildfire ignitions across a 1,000 km² landscape within just ten minutes. The finals will take place in Nenana, Alaska, where teams will demonstrate their technologies in realistic wildfire response scenarios.

Being selected as a finalist places AURA Foresight among a small group of innovators developing the next generation of tools to help communities, land managers and firefighters respond to wildfires faster and more effectively.

As wildfires become more frequent and severe worldwide, AURA Foresight is developing technology that helps stop fires before they become disasters. The team’s autonomous wildfire intelligence and intervention system is designed to detect ignitions within minutes and coordinate a rapid response while fires are still small and manageable.

Detect Early. Verify Quickly. Intervene Immediately.

AURA Foresight combines fixed sensors, artificial intelligence and swarms of autonomous flying robots into a seamless detect–verify–act capability.

Rather than fighting large wildfires after they have taken hold, the system is designed to identify potential ignitions, verify threats and intervene at the earliest possible stage, when a small spark can still be safely contained before becoming a major incident.

The platform continuously monitors landscapes using optical and thermal sensors, automatically identifies potential ignitions using AI, dispatches flying robots to verify threats, and coordinates rapid intervention before small fires can grow into destructive wildfires.

Unlike many wildfire technologies that rely on specialised infrastructure or proprietary hardware, AURA Foresight has been designed from the outset to be practical, affordable and easy to deploy.

Key advantages include:

• Out-of-the-box deployment with minimal setup requirements.
• Adaptability to any environment, from forests and remote wilderness to critical infrastructure corridors.
• Scalability from a handful to dozens of flying robots, depending on operational needs.
• Compatibility with different aircraft and robot platforms, avoiding vendor lock-in.
• Built on commercially available technologies, making advanced wildfire protection accessible and achievable for organisations of all sizes.

The result is a system that can help agencies move from passive monitoring and delayed response to proactive, automated intervention.

Developed With Firefighters, For Firefighters

AURA Foresight’s approach is shaped by close collaboration with firefighters and emergency response professionals. The consortium works directly with Lancashire Fire and Rescue Service in the United Kingdom and with wildfire response experts in Canada and Australia to ensure the technology addresses real operational challenges and integrates effectively into existing emergency response workflows.

This combination of frontline expertise and advanced robotics helps ensure that the technology remains focused on supporting firefighters and enhancing their capabilities in the field.

“It’s amazing what can be achieved when great people come together around a shared purpose,” said Dr Georgios Tzoumas, Team Co-Lead of AURA Foresight and Research Fellow at the University of Bristol.

“Our team has been working on autonomous wildfire detection and intervention technologies for more than six years. Reaching the XPRIZE finals is an exciting milestone because it feels like the solution is finally within reach. By combining AI, swarm robotics and close collaboration with firefighters, we’re showing that it’s possible to tackle wildfires while they’re still small, before they grow out of control. We can’t wait to demonstrate our technology in the field in Alaska.”

A Global Consortium Combining Expertise from the UK and Australia

AURA Foresight is a unique international collaboration bringing together AURA in the United Kingdom and Fire Foresight in Australia.

The consortium includes the University of Bristol, Bristol Robotics Laboratory, the University of Sheffield, Lancashire Fire and Rescue Service, Fire Foresight, SkyFly Drones, Southern Denmark University, Manchester University, Indicium Dynamics, Robotic Cats, Taz Drone Solutions and Little Place Labs.

Fire Foresight CEO Rob Vernon commented, “Our approach with the XPRIZE competition has been to bring all the pieces of the wildfire jigsaw puzzle together, and that capability only truly scales when it’s done with partners from around the world and on the ground. In Australia, we have been building the largest network of bushfire detection capabilities and the partnerships that we’ve formed, and continue to form, through this competition allow us to focus our attention on the autonomous mitigation and suppression capabilities, so desperately need to address the challenges faced by wildfire and climate change”.

Together, these partners combine world-leading expertise in swarm robotics, autonomous systems, artificial intelligence, computer vision, aerial robotics, wildfire operations and field deployment.

By bringing together researchers, engineers, firefighters and technology innovators from across two continents, AURA Foresight is demonstrating how international collaboration can accelerate solutions to one of the world’s most pressing climate challenges.

As destructive wildfires continue to increase in frequency and severity around the world, AURA Foresight’s progression to the XPRIZE Wildfire finals highlights a new approach to wildfire management: tackling fires in their earliest moments, before they can grow out of control and threaten communities, ecosystems and critical infrastructure.

Detect early. Verify quickly. Intervene immediately. Stop wildfires before they become disasters.

Unmanned aerial vehicles (UAVs), commonly known as drones, are now widely used for various purposes, ranging from filmmaking and aerial photography to industrial inspection, precision farming and reaching obstructed areas during emergency response missions. While many existing drones can move swiftly in their surroundings and circumvent large obstacles, most still struggle in cluttered environments. In addition, they are often unable to execute maneuvers that would allow them to safely pass through small gaps or reach secluded areas.