By Guest: Arti Raman
Company: Portal26
Show: The Agentic Enterprise
Topic: AI Governance
Billions of dollars are flowing into generative AI, yet the vast majority of enterprise AI initiatives never make it past the proof-of-concept stage. According to figures cited by Arti Raman, CEO of Portal26, only 4% of AI projects are currently delivering positive ROI — and the POC-to-production conversion rate sits between a dismal 5% and 15%. The question isn’t whether enterprises are committed to AI. They clearly are. The question is why that commitment keeps failing to produce results.
📹 Going on record for 2026? We're recording the TFiR Prediction Series through mid-February. If you have a bold take on where AI Infrastructure, Cloud Native, or Enterprise IT is heading—we want to hear it. [Reserve your slot
Raman has a clear answer: companies are picking the wrong projects, in the wrong way.
“The projects being conceived are coming out of top-down brainstorming,” she explained in a recent conversation with TFiR. “Companies bring in consulting firms, leadership teams get together, and everyone has really good ideas. But the hit rate for those ideas is literally five out of 100.”
Meanwhile, the employees actually doing the work — developers, customer support teams, creative staff — have already figured out where AI delivers value. They’re using it every day, often through unsanctioned tools. The problem is that organizations have no way to capture those demand signals and turn them into informed investment decisions.
That gap is exactly what Portal26 was built to close.
Three Pillars: Visibility, Security, ROI
Founded in 2019 with an initial focus on data security, Portal26 expanded into a full AI adoption platform in 2023. The company’s approach rests on three interconnected pillars: visibility, security, and value realization.
Visibility is where everything starts. “We tell our customers: we just turn the lights on,” Raman said. “You’re in a dark room, you don’t know what’s happening — you turn the lights on, you can see.” Within weeks of deployment, Portal26 begins surfacing how AI is actually being used across an organization — which tools employees are reaching for, what they’re using them for, and where the highest activity is concentrated.
Security is the second pillar, and it takes on new dimensions in the AI era. Unlike previous technologies, AI models retain context and can be further trained on data fed into them — making unsanctioned usage a serious risk. Portal26 monitors both sanctioned and personal AI tools, intercepts sensitive data before it leaves the organization, and can redirect users toward approved workflows in real time. “It will stop it from going out if it catches something that shouldn’t be,” Raman explained, noting that the platform can also educate users in the moment — explaining policy and offering compliant alternatives rather than simply blocking behavior.
The third pillar — ROI — is where Portal26’s newest module, the AI Value Realization Platform, comes in.
License Intelligence and the Demand Signal Problem
One of the most striking discoveries Portal26 has made with its customers involves a fundamental mismatch between what organizations license and what employees actually use. In one example Raman shared, a company was spending $60 per engineer per month on two licensed AI coding tools — yet Portal26’s monitoring revealed that developers were visiting ChatGPT and DeepSeek tens of thousands of times per month to write production code.
“The question is: why are developers going to ChatGPT and DeepSeek to write production code when they have access to these tools?” Raman said. The answer, in most cases, is that the licensed tools weren’t actually solving the problems developers needed solved. The investment was real. The fit wasn’t.
Portal26’s License Intelligence module surfaces exactly these mismatches — showing which licensed tools are being used, which public alternatives are competing for user attention, and what the actual demand pattern looks like. From there, the Value Realization Platform layers in Risk Value Analytics, which maps high-usage behaviors against their security risk profile, helping organizations prioritize which use cases to build out, secure, or retire.
The result is a pipeline of validated, high-ROI use cases that can feed directly into agent studios and AI development platforms — ensuring that what gets built is something employees will actually use.
From Purgatory to Production
Raman is optimistic about where the industry heads from here. In Portal26’s own customer base, she reports dramatically higher POC-to-production conversion rates compared to industry averages, and early results with the Value Realization Platform are showing meaningful ROI improvements. Across the broader market, she expects the positive ROI rate to climb from today’s 4% to somewhere between 25% and 35% within a year, with POC conversion rates moving from 10-15% to 50-60%.
“CFOs have been pouring money into it for two and a half years,” she noted. “Now they’re asking hard questions: who’s actually using it? Who’s not?”
For organizations ready to find out, Portal26 offers a low-friction starting point. The platform deploys in roughly 30 minutes and can deliver a meaningful first assessment within hours or days. “We encourage people to just come and turn the lights on,” Raman said. “See what’s happening in your organization — and from there, you decide what to do about it.”
