Videos
Cursor on Scaling Enterprise AI Adoption
Insights
- Most enterprises struggle to scale AI because they lack clear strategy, governance, risk controls, operational redesign, and measurable business outcomes.
- AI adoption succeeds when organizations redesign the entire software development lifecycle around AI-first workflows rather than layering AI onto existing processes.
- Enterprise AI governance must evolve beyond policy documents toward real-time auditability, observability, and operational accountability.
At the AI Horizon event in Houston, Mit Majumdar, EVP and Global Head of Services at Infosys speaks with Chris Diaz, Field Engineer at Cursor, about the challenges enterprises face as they move from AI pilots toward scaled adoption. The discussion highlights five major gaps slowing enterprise AI progress, including missing strategic clarity, weak governance, undefined risk frameworks, difficulty measuring business impact, and outdated operating models that fail to support AI-native workflows. Chris Diaz shares how organizations such as NVIDIA are redesigning their software development lifecycle around AI-first engineering practices, integrating agentic coding tools directly into planning, development, QA, deployment, and monitoring processes. The conversation also explores how enterprises must rethink leadership, governance, and organizational change management to operationalize AI successfully at scale across highly regulated industries.
Mit Majumdar:
We know Enterprise AI is such a big and important topic in today's industry scenario. And to have more discussion on this topic, we have Chris Diaz from Cursor, who is a field staff engineer. Chris, thank you for having this conversation with us.
Chris Diaz:
Thanks for having me, Mit. I'm excited to be here today.
Mit Majumdar:
So first question, Chris, just give us a little bit of a sense of what is your role and what do you do within Cursor?
Chris Diaz:
Yeah, so at Cursor, I'm a field engineer. And what that means is I work closely with our largest and most strategic customers to evaluate, implement, and adopt agentic coding solutions. And I love this role because I get to work with so many engineering teams and help them change the way that they're delivering software.
Mit Majumdar:
One question that all of us are grappling with today is that this whole era of pilots, implementing pilots using AI to scaling its adoption in large scale enterprises. From your experience, what gaps do you see that is inhibiting the growth from pilot to large scale adoption?
Chris Diaz:
I think there's five gaps that I've observed with all of the enterprise customers that I've been working with.
And the first starts at this pilot phase, where there's missing strategic clarity, right? AI everywhere is not a strategy. That's a hope. And so organizations need to identify where can AI provide a durable advantage? And so what's really important here is specificity, not breadth. And then that's the first one.
The second one that I see is risk appetite. A developer running AI in an isolated sandbox has a very different risk profile than an autonomous AI agent running in a CI pipeline making production changes. And so customers oftentimes aren't thinking about the risk appetite and creating that tiering when they're deploying AI.
The third one's closely related around controls and governance. It's very interesting to see how many enterprises aren't able to answer basic auditing questions. And so if I ask a customer, what did this AI agent do last Thursday, they should be able to show me an audit log with the user that invoked it, the prompt, the model that was used, the actions that were taken. Oftentimes, they just show me a PDF of who owns what, and that's just governance theater.
The fourth area is understanding impact and measurement. Many enterprises are struggling to connect the adoption metrics to actual business impact, like change rates, defect rates, margin impact, revenue impact. And so they oftentimes are able to get to certain engineering impacts like PR cycle time, but still struggling to connect that to business value.
And then the fifth gap I commonly see is around operational strategy. Are we just throwing AI on top of an existing process? Or are we actually rewriting the playbook? And oftentimes companies want to just drop it on top and are expecting amazing results. But this is the same thing we saw with the migration to cloud that Infosys, I think, saw with a lot of customers. You can't just lift and shift applications. You might get marginal benefits, or you might be creating tech debt. And so having an operational strategy around how is this going to change the process? Are the incentives aligned? Is leadership aligned to lead AI assisted teams? That's another area where I'm seeing a gap.
Mit Majumdar:
Excellent. So that's like a five-point strategy that you're talking about where you're seeing gaps. And those gaps have to be circumvented to be able to see a large-scale adoption.
Chris Diaz:
I think if we want to see large-scale adoption that's actually providing results, these are things that customers need to think about and have a defined strategy to actually realize the benefits that AI can provide.
Mit Majumdar:
So on that, can you share a large case study of… in Cursor where you have been able to circumvent some of these things and customers have been able to address them and hence you have seen a large scale adoption.
Chris Diaz:
One that comes to mind is NVIDIA. And so NVIDIA has rolled Cursor out to 30,000 developers now that are using Cursor daily. Last week at GTC, Jensen Huang told the world that Cursor is his favorite agentic coding tool.
And the reason they've found success, well, the success that they've seen is they're seeing a 3x amount of code that's being committed from before, while their defect rate has remained flat. The reason that they're finding and seeing this success is because they've defined a strategy that addresses those five gaps. And the way that they've implemented this is they didn't just hand out a coding tool to their developers, but they actually have integrated that tool into their SDLC. So from planning to design to development, to QA, and then finally to deployment and monitoring.
And so now they don't have just Cursor sitting on top of an existing process, but they've redefined their SDLC to be an AI-first SDLC, which is allowing them to see these results at scale.
Mit Majumdar:
So we at Infosys are really excited about this partnership that we have announced with Cursor. And one of the primary objectives is large scale adoption for our customers and bringing value to them. From your point of view, how do you think this partnership will help customers going forward?
Chris Diaz:
Yeah, I think this partnership is going to bring a lot of value to both of our customers because those five points where we're seeing gaps, those are actually core competencies of Infosys that Infosys engineers excel at, right? Infosys has a history of helping drive change management in large organizations and addressing these gaps where we understand what is the strategic impact and the strategic reason for implementing here. How do we define risk controls, governance, but then also how do we help realize the impact here? And then of course driving that organizational change and making sure that the operational strategy aligns with the goals that we're trying to achieve. And also the domain aspect.
And then absolutely. So the other thing is, Infosys has these rich experiences with customers across financial institutions, healthcare, technology companies, and every other Fortune 500 industry. That domain expertise helps enterprises adopt AI at scale to get these results that are often being advertised.
Mit Majumdar:
Chris, this was a fascinating conversation. Thank you so much for taking the time and talking to us. Lovely meeting you again.
Chris Diaz:
Yeah, thanks for having me, Mit. It was great to be here. Thank you.
