|

Agentic AI showcased at HPE Discover 2025

⏱️ Estimated reading time: Approx. 7 minutes (7 min read)
HPE Discover 2025에서 보여준 Agentic AI - image 4

On June 23, 2025, HPE successfully hosted Discover 2025 in Las Vegas.

On this day, HPE CEO Antonio Neri unveiled GreenLake Intelligence, a groundbreaking framework for applying agentic AIOps technologies across hybrid IT infrastructure.

And soon, a new cloud-native network management system (NMS), HPE Networking Central, will be available. Agentic Mesh featureThis will be equipped with HPE's GreenLake Intelligence. First specific area serviceIt will be.

This agentic message Fully autonomous driving networkIt will greatly advance our steps towards that goal.

Our AI for Networking The Journey: The Story So Far

Networks are never predictable systems.

From gateways and switching to the last meter of 6GHz wireless connectivity, your network is a distributed IT system with endless configurations and possibilities.
That's why AI for NetworkingIt's no surprise that it delivers ROI value across your hybrid IT infrastructure.

HPE Discover 2025에서 보여준 Agentic AI - image 30

That's why HPE has invested heavily in AI innovation in HPE Aruba Networking Central over the past five years.
Let's briefly look back at the process so far.


Episode 1 (2019 and beyond) – Classification and Prediction AI

HPE Discover 2025에서 보여준 Agentic AI - image 7

The value of AI depends on the quality, scale, and variety of the data it uses to make inferences.

HPE Aruba Networking Central is now Over 3 billion client terminalsProviding services to Over 6 million network devices, and generates very large big data streams.

HPE Discover 2025에서 보여준 Agentic AI - image 11

Working in high-performance data pipelines Various unsupervised and supervised learning modelsWe continuously monitor and analyze over 1 trillion network telemetry data points every day.

such Non-generative classification and prediction modelsis essential to the AIOps stack.

Because modern LLMs are not designed to process such massive and continuous time series data in real time.
HPE leveraged non-generative technologies across a variety of areas, including switches, WAN, 6GHz, WPA3, and Green AP optimization.
Through this More than triple the network optimization insights generated by AI.I did.

Episode 2 (2024 and beyond) – The rise of GenAI, LLM, and Transformers

HPE Discover 2025에서 보여준 Agentic AI - image 12

HPE has integrated enterprise-grade, privacy-focused generative AI (GenAI) technology into its AI search engine.
It is one of the first network management systems to significantly improve its natural language understanding and response capabilities for search and document summary queries.

The first LLM implementation is Online and offline RAG (Research Augmented Generation)1 FrameworkIt is used.
Thanks to this, the hallucinations that commonly occur in general LLMs have been reduced.

Additionally, we have safeguards in place to protect sensitive customer data, such as personally identifiable information (PII/CII).2 We implemented a transformer that internally learns the technology.

HPE Discover 2025에서 보여준 Agentic AI - image 16
RAG (Augmented Search Generation)

Generative AI technology has enabled significant advancements in the accuracy and usability of search.
Recently, LLM has been applied to increase the understanding of security vulnerabilities (CVE). Enhanced predictive firmware recommendation featureI did.

Additionally, we've improved our endpoint security classification capabilities.

  • HPE Discover 2025에서 보여준 Agentic AI - image 13
  • HPE Discover 2025에서 보여준 Agentic AI - image 14
  • HPE Discover 2025에서 보여준 Agentic AI - image 15

Episode 3 (Today) – Agentic Mesh Debuts

HPE Discover 2025에서 보여준 Agentic AI - image 17

Since 2016, HPE has been evolving to introduce a human-centric autonomous driving network.
The rapid advancements in AI over the past two years have brought us much closer to achieving that goal, and new autonomous driving capabilities are soon on the horizon.

Agentic systems require specific goals that can be achieved through autonomy.

Agentic AI implemented in HPE Aruba Networking Central provides accurate root cause analysis and guided or automated problem resolution.
Agentic AI complements the limitations of existing LLM-based generative AI to support autonomous judgment and processing.

  • HPE Discover 2025에서 보여준 Agentic AI - image 23
  • HPE Discover 2025에서 보여준 Agentic AI - image 24
  • HPE Discover 2025에서 보여준 Agentic AI - image 25

We've integrated AI agents into HPE Aruba Networking Central that independently perform specific specialized tasks (e.g., network configuration generators).

These new AI agents have specific properties:

  • inference (Reasoning): To enable the network to make its own judgments and decisions, Utilize the industry's best LLMs (Large Language Models)Here are the data scientists Additional learning of network-related dataIt was made so that you can make more sophisticated judgments.
  • Orchestration (Orchestration): A new autonomous multi-agent orchestrator (MAO) is introduced.
    MAO is like Like a field commander Leverage advanced reasoning capabilities to analyze and act on real-time network data, configuration information, or existing knowledge. Understand, coordinate, and direct specialized sub-agentsdo.
    MAO to each sub-agent Role-based authorizationThus, it activates the relevant sub-agents only when necessary.
    With this A structure in which various experts collaborateIt guides the entire agent system to reach accurate conclusions efficiently and effectively.
  • New human interface (New Human Interface): Agentic AI should not be thought of simply as an advanced conversational interface based on LLM. The ultimate goal is Autonomous systems with human-centered designBecause it is to create, the solution is Network-specific GreenLake Copilot interfaceIncludes:.
    This new Multimodal conversational AI companionOn behalf of MAO You can initiate a conversation with a real-life operator, or you can have the operator answer your questions.

Networking co-pilots are constantly interacting with users and showing them the flow of their thoughts.
This allows people to directly see the reasoning behind the conclusions, recommendations, or actions MAOs make.

HPE has successfully integrated all of these technology elements and implemented them as 'Agentic Mesh' in HPE Aruba Networking.

Solving Roaming Issues with Agentic Mesh

The system not only efficiently analyzes massive network data, but also provides AIOps (AI-powered IT operations) capabilities that perform advanced inferences on their own about complex network and security situations.

Are your network and IT teams ready for agentic autonomy?

HPE believes so, but actual implementation will vary depending on customer circumstances.
Many clients tell us, "IT professionals are increasingly becoming multi-disciplinary, making it difficult to specialize in a specific area.".

Reflecting this reality, Agentic and Copilot featuresAs if Someone who can work and talk with you like a new colleague on the teamIt is designed with .
The most important thing is that they Provide real value to employeesThe point is that it will be done.

Furthermore, HPE's networking expertise will soon be GreenLake IntelligenceAs part of Interoperability with other agentic meshes, not just networksIt will encompass the entire hybrid IT stack.

The agentic era has finally arrived on the network!

HPE Discover 2025에서 보여준 Agentic AI - image 29

HPE is not only presenting its vision for Agentc AI, but also has plans to bring it to market in the near future. This cutting-edge functionality is deeply integrated into our NMS (Network Management System) product, Central.I will be able to show you what I have done.

HPE Aruba Networking's New Agentic AI Meshis networking with GreenLake Co-Pilot Expected to be available in HPE Aruba Networking Central environments starting in late summer 2025.no see.


  1. Retrieve required information externally and add it to the prompt before generating an answer. ↩︎
  2. Designing AI systems to operate safely, ethically, and within their intended scope. ↩︎

Similar Posts

  • HPE Aruba Networking Wireless LAN Portfolio

    ⏱️ Estimated reading time: Approx. 5 minutes (5 min read) Aruba began its business in 2002 as a company specializing in wireless networks under the name Aruba Networks. Leading up to its merger with HPE, Aruba's wireless network solutions demonstrated industry-leading technology and performance. HPE Aruba Networking WLAN is intelligent, secure, and… through AI-based mobility and enterprise-grade design

  • Where will HPE Aruba be at RSAC 2024?

    ⏱️ Estimated read time: Approx. 6 minutes (6 min read) Due to generative AI and evolving threats, security teams must rely on networks as a new tool for Zero Trust security. At the RSA Conference 2024, HPE Aruba Networking unveiled new AI-based network security… that supports security teams in protecting valuable AI assets from Edge to Cloud…

  • Why Wi-Fi Authentication Matters

    ⏱️ Estimated reading time: Approx. 5 minutes (5 min read) We are deeply immersed in the fun of testing out new technology by shouting, “Alexa, turn on the Christmas tree” or “Ok Google, turn on the Christmas lights.” However, while this command perfectly worked Christmas tree lights that were fully compatible with Alexa, it didn't work so smoothly with Google…

  • Gartner Magic Quadrant Leader for 19 Consecutive Years

    ⏱️ Estimated reading time: Approx. 2 minutes (2 min read) Gartner® has once again recognized HPE (HPE Aruba Networking) as a leader in the Magic Quadrant for Enterprise Wired and Wireless LAN Infrastructure, making HPE the only company to have received continuous recognition since the report was first published in 2006. Gartner [identifies] the enterprise wired and wireless LAN infrastructure market with users and…

  • HPE Aruba Networking meets the F1 Mercedes team in an innovative way

    HPE has built a data-driven partnership with the Mercedes-AMG PETRONAS team to continuously improve vehicle performance. HPC and high-speed network solutions are used to reduce simulation time, deliver real-time data for continuous improvement, and ultimately provide a competitive advantage in racing.