Speed, trust and interoperability — artificial intelligence has officially entered its agentic era. While many enterprises still fixate on narrow productivity gains, boardroom discussions are moving toward finding an AI platform that can support agents at scale. 

Fabrix.ai’s Bhaskar Krishnamsetty, Rached Blili and Shailesh Manjrekar talk with theCUBE about building and scaling their agentic AI platform.

Fabrix.ai Inc. has presented a clear vision for this future. The company introduced an operational intelligence platform built on data and automation fabrics, supported by a partner ecosystem that includes power players such as Cisco Systems Inc. and IBM Corp. But behind the demos and alliances, the company’s leadership keeps returning to a single theme: Agentic AI only matters if it closes the gap between AI projects and tangible business outcomes, according to Shailesh Manjrekar (pictured), chief AI and marketing officer at Fabrix.ai.

“We are really at an inflection point with agentic, right? Every enterprise, small or large, wants to be an autonomous enterprise,” Manjrekar told theCUBE during the event’s opening session. “However, what we are seeing is there is what I call an AI value gap. There are a lot of these AI projects happening, but really how much they contribute to … an outcome is [what] enterprises are struggling with.”

Manjrekar spoke with Bob Laliberte, principal analyst at theCUBE Research, at the “Agentic AI Unleashed: The Future of Digital & IT Operations” event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the current AI value gap many enterprises are struggling to bridge — and how they are positioning their AI platform at the core of the agentic conversation. (* Disclosure below.)

Here’s the complete video interview with Manjrekar:

Here are three key insights you may have missed from theCUBE’s coverage:

Insight #1: For a sustainable agentic AI platform, speed can’t outpace observability.

If AI-powered agents are to move into real operations, they can’t be relegated to recommendation engines. They’ll need to prove their worth while providing visibility across technology stacks, acting quickly without neglecting an audit trail. Fabrix.ai differentiates its platform by positioning its fabrics as the link between alerts and action in complex environments, according to Satyan Raju, chief development officer of Fabrix.ai. 

“We saw this with a customer who had an outage on a business-critical production application,” he explained during an interview with theCUBE. “Our [root cause analysis] agent identified a flapping network switch as a root cause and provided insights for immediate remediation, even though the logs were not fully enabled across all devices. That speed was possible because of our full-stack cross-domain correlation powered by our AI and data fabric.”

Measuring that kind of impact means treating AI as a production system, not a black box. The most advanced customers now track the health of their agentic AI with the same rigor they apply to infrastructure, according to Tejo Prayaga, vice president of product management at Fabrix.ai Inc. That includes both hard metrics and human factors, such as how teams actually use and trust the outcomes.

“Enterprises want the same kind of visibility into AI that they would expect from any other production system, and these are the metrics like model accuracy, inference costs, latency and success rates,” he said. “Now it’s all about understanding how efficiently and transparently AI is performing … It comes down to productivity and the team expertise and experience. Here we are seeing much better coordination across teams because they are operating with that shared context.”

During the event, host Zeus Kerravala, principal analyst at ZK Research, joined many of the sessions — including theCUBE’s complete video interview with Raju and Prayaga. This interview also features insights from Venkat Raghav, chief technology officer and global offering leader for AIOps, information technology automation and DevSecOps at IBM Consulting:

Insight #2: An agentic AI platform needs an open, interoperable ecosystem.

All real-world deployments of any AI platform will span networks, clouds and governance systems, meaning it must be designed for interoperability from the outset. Fabrix.ai positions its partner strategy as essential for becoming a credible solution for global enterprises.

In a key insight session, partners such as Cisco’s Papi Menon spoke with theCUBE about the Fabrix.ai AI platform and partner ecosystem.

“Cisco provides the secure AI-ready infrastructure and observability fabric. Splunk delivers the analytics and correlation layer across the data. The way I see [Fabrix.ai], it adds the agentic intelligence that turns those insights into autonomous remediation and optimization,” said Mohannad Abuissa, managing director of solution engineering and chief technology officer for the Middle East, Africa, Turkey, Romania and the Commonwealth of Independent States at Cisco. “Together, the whole stack enables customers to move from visibility to action.”

On the standards side, Cisco’s Outshift Agency project is working to ensure that multi-agent systems don’t fragment into incompatible islands. For agentic AI to scale across enterprises, the “internet of agents” must be rooted in open, neutral protocols. That’s the basis for getting different agent frameworks to talk to each other, and for partners such as Fabrix.ai to plug into a common ecosystem.

“It’s been a collective effort,” Papi Menon, vice president of product management and chief product officer of Outshift by Cisco, told theCUBE. “We realized that anytime we talk about the internet of agents, if you’re talking about open interoperable standards for agentic collaboration and communication, that’s something that needs to be done as a collective, because everyone has to adopt it. It has to be a neutral standard.”

Here’s theCUBE’s complete video interview with Abuissa, Menon and Manjrekar, with further insights from Andy Thurai, field chief technology officer of Splunk Inc., another key partner in Fabrix.ai’s ecosystem:

Insight #3: Automation should upgrade human roles, not replace them.

While any agentic AI platform promises to automate workflows, it also requires organizations to reconsider the division of labor between humans and digital coworkers. Fabrix.ai and its partners emphasize that agentic AI should remove repetitive tasks from operations, enabling human talent to focus on strategy.

Fabrix.ai’s Raju Penmetsa and Shailesh Manjrekar talk with theCUBE about AI agents and operational efficiency.

“If you take AIOps … where one of our customers was the largest service provider in the world and [their] L1 is completely automated to the point where L1 people are doing L2 [tasks] … [the] people got the time to upskill themselves to support L2, and that is a transformation for an organization if you are in services,” Raju Penmetsa, chief technology officer of Fabrix.ai, told theCUBE. “When you are doing complex things, it can give you human in the loop, so that the system gives them the ability to make a decision … rather than crunching all the numbers.”

In the agentic AI era, enterprises must be aware that IT shifts from performing tasks to managing a digital workforce. Operations professionals remain essential, but their daily responsibilities move closer to strategy and business design as agents handle repetitive work, allowing humans to focus on higher-level initiatives, Manjrekar explained.

“That’s what they’re going to be relegated to for the most part. I mean, again, it’s not from an operational perspective,” he said. “Now, they can choose to do a lot more value-added things when it comes to strategic operations.”

Here’s the complete video interview with Manjrekar and Penmetsa:

(* Disclosure: TheCUBE is a paid media partner for the “Agentic AI Unleashed: The Future of Digital & IT Operations” event. Neither Fabrix.ai, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE