How NextDoor's CIO Is Trading SLAs For 'Experience Level Agreements' To Match AI Demands

While enterprises are rushing to adopt AI, few are turning it into real, measurable value. The promise of ROI is everywhere, but success in AI isn't about ambition, it's in the execution. So much attention has focused on 'just getting started' that as enterprises are now pushing projects into production, they're recognizing the tools used to measure success are relics of a bygone era and miss the mark. For decades, the industry has relied on Service Level Agreements (SLAs)—rigid, technical metrics like uptime, response time, and ticket resolution. In today's environment, where budgets, outcomes, and ROI are under a microscope, these metrics are falling critically short.

The limitation of SLAs is that they don't capture user experience, productivity, or the actual business value delivered. They only measure whether a service meets a technical threshold. In short, SLAs ask, 'Did the system work?' but the far more important question for AI is, 'Did it make the user's experience better?' This fundamental gap has left leaders struggling to justify their investments and prove that AI isn't just a cost center. The answer, according to some experts, lies in a radical shift in measurement, from service to experience.

CIO News spoke with Shyam Bhojwani, the Chief Information Officer & Head of Security (CISO) at Nextdoor, to see how he's drawing on his experience leading enterprise technology and automation to implement new frameworks to measure and track success of AI pilots and projects. Bhojwani argued that to unlock the true value potential of AI, leaders must abandon outdated metrics and embrace a new framework designed for the modern enterprise: The Experience Level Agreement (XLA).

• From service to experience: "With XLA, it's focused explicitly on how am I improving the experience? Am I reducing context switching? Am I giving information quickly? Am I being proactive, not reactive? That's a real metric enterprise CIOs should focus on," Bhojwani said. "XLAs were brought up very loosely last year, but this year AI is out of pilots and into production serving as a forcing function for teams to measure results actively."

• Everyone is a CEO: This new framework and the metrics it's tracking aren't just about better data; it's about fostering a new culture of empowerment. "Everyone should operate like a CEO. Your AI is doing most of the work," Bhojwani stated. "We all are CEOs of our business process. We are owners of our end-to-end things we do."

For leaders inspired to make the shift from SLAs to XLAs, Bhojwani's advice is to be both patient and pragmatic. This is not a one-size-fits-all solution, but an iterative process that requires customization and a willingness to start small.

• Start small, iterate often: "Pick one team or department and measure for a week, a month, and a quarter to see if the structure makes sense. If it does, deploy it across the company," Bhojwani advised. "That's how you can show that your AI investments are paying off. It's not just about dollars saved, but about how you are enabling your employees. That's the big question to be answered."

• The red flag metric: The key to knowing if a pilot is working is to measure context switching. If employees are spending less time toggling between applications, the XLA is improving outcomes and helping drive value. However, if adoption is low, this can be a critical warning sign to watch for according to Bhojwani, "You need to investigate and have conversations, because it could be as simple as an enablement issue; they don't know how to use it."

Ultimately, adopting an XLA framework is about ensuring that massive investments in AI are not just lighting cash on fire, but are generating tangible returns in productivity and employee empowerment. It requires a fundamental change in mindset, viewing AI not as a static tool but as a dynamic, proactive force that is reshaping the nature of work itself.

• Accelerating empowerment: "Automation was sort of phase two of industrial revolution. AI is phase three," Bhojwani concluded. "But the big difference is AI is proactive. It can change according to your business, and so that sort of proactive mindset is super important... it's not static, it's dynamic. It can change according to your business very quickly so that actually empowers you."

Evaluating metrics is only possible by having a strong and disciplined framework to launch and orchestrate AI across the enterprise. Often, the rush to deploy AI can lead to chaos and the next generation of tech bloat. Bhojwani warned against AI sprawl—a state where hundreds of unmanaged, redundant intelligent agents are created across an organization, creating technical debt and security risks. To prevent this, he has implemented a strong governance model at Nextdoor focused on bringing order to the agentic ecosystem.

• The AIR up there: As enterprises start launching pilots or even move projects to production, you don't want to create 100 agents without some form of organization. "Whatever agents you're creating need a good naming convention, should be cataloged, and map to a taxonomy of specific business units or workflows," Bhojwani advised. "We used to have user access reviews, or UARs, and now we have AI Agent Reviews, or AIR."

• Do Not Pass Go: To assist with eliminating added work on the backend of the process, Bhojwani shared how they've added banners on their agentic platform that say, 'Before creating an agent, do a search.' According to Bhojwani, the most important part of determining when to create a new agent is surprisingly tech-free. "You need to initiate conversations with your peers and leadership, especially for cross-collaboration teams, and ask 'I'm trying to build this agent, what are your thoughts?'"

• Last Line of Defense: Bhojwani's strategy for preventing sprawl is a three-pronged defense combining strong governance posturing, proactive guardrails, and reactive cleanup. He shared, "We always include human-in-the-loop by conducting a thorough review every week to see how many new agents were created and what apps they are hitting. If there are repetitive or overlapping agents we go and talk with the teams who've created them to determine how we can drive efficiency."

This new framework for AI is not just a technical challenge; it's a cultural one. Bhojwani stressed that successful AI adoption requires breaking down organizational silos and getting universal buy-in from the board room to the individual contributors. From his perspective, the goal is to treat AI as a unified, cross-functional capability, not a series of isolated projects.

• United front: "The role of a CIO or IT leader is becoming more important because everyone wants to leverage AI, and we are the gatekeeper, the implementer, and the facilitator," he said. "It's important to get buy-in from the very top, from the board and CEO level. Don't create siloed AI; AI is one thing across the company, so that cross-functional buy-in is also important."

• A two-way street: This cultural shift also involves reframing AI as a tool for empowerment, not a threat to job security, and establishing shared accountability. "People who use AI smartly, effectively, cross collaborate, I think they're going to succeed," Bhojwani noted. "It's my job to enable them, but they also need to take action... it's a two way street equal engagement from both sides."

By combining governance, cultural alignment, and outcome-focused metrics, Nextdoor is demonstrating how enterprises can move beyond hype to generate measurable value from AI. For CIOs and IT leaders looking to lead AI adoption, the lesson is clear: focus on execution, measure experience, and empower people, not just technology. Bhojwani sums it up best, "AI is a democratizer. It’s not about hierarchy anymore — it’s about who uses it best. That’s the new competitive advantage."