In this exclusive interview with Swapnil Bhartiya at TFiR, Jill Adams, VP of AI Product and Experience at Copado, discusses how the Salesforce ecosystem can bridge the gap between AI-native development tools and enterprise governance requirements through AgentOps and the Agentia platform.

The Governance Crisis in AI-Native Development

AI agents are transforming software development velocity, but they’re exposing a critical vulnerability in traditional DevOps workflows. While tools like vibe coding accelerate productivity by orders of magnitude, they operate without awareness of organizational context, compliance requirements, or release governance. This creates a paradox: teams move faster but with less control.

Q: What gap exists between AI coding tools and enterprise delivery pipelines?

Jill Adams: “Today, AI tools are doing all the work, but in most cases they have no idea who we are. They don’t know your Salesforce org. They don’t know your metadata, your release pipeline, or your compliance requirements. All they do is generate code, test ideas, but in vacuum. And when something breaks in production, there is no audit trail, no governance, and actually no one in control. That gap between AI capability and enterprise reality is exactly what Copado is closing with Agentia.”

Q: How does AI-generated code volume create new risks?

Jill Adams: “What we have seen is that with the introduction of especially vibe coding tools, for example, that 10X, 100X, the productivity of developers, all of a sudden we’re getting more code, but that creates more risk because more code is being introduced, more merge conflicts. And so the weight of that code and the weight of the changes is now almost too much for individual people to actually be managing that process on their own. And that’s why we firmly believe that if you’re going to be introducing vibe coding, if you’re going to be using agents in your development process, you also need agents as part of your DevOps process. And that’s what we’re calling agent ops. And that’s going to help remove that bottleneck of the human being the bottleneck in that process.”

Copado’s Foundation: 13 Years of Salesforce-Native DevOps at Enterprise Scale

Copado’s entry into the AgentOps space isn’t a pivot—it’s an extension of proven infrastructure. The company’s 13-year history delivering Salesforce-native DevOps solutions for the largest, most complex enterprise environments provides the governance scaffolding that AI agents need to operate safely in production.

Q: What is Copado’s foundation and how does it position the company for AI-native development?

Jill Adams: “Copado has been a Salesforce native DevOps solution going back 13 years. And so that is really our foundation is that we have proven scale and governance across the largest enterprises that use Salesforce, complex environments, complex compliance requirements, complex governance needs. So that’s really our foundation. And over the last few years, what we have noticed is that AI is becoming more an imperative, obviously, with the introduction of AI native development tools. And so what we’ve noticed is there’s actually a gap today between being able to leverage those tools and having safe and governed releases and bringing those agents into production. There’s a gap there. So what our goal is from Copado is to bridge that gap and be the bridge from DevOps to what we’re calling agent ops.”

From DevOps to AgentOps: Why Traditional Pipelines Break Down

The transition from DevOps to AgentOps represents a fundamental shift in how organizations manage software delivery. Traditional DevOps workflows were designed for human-driven commits, pull requests, and releases. Agent-driven workflows operate at different velocities, with non-deterministic behavior patterns that require new governance models.

Q: Why does AI break the traditional DevOps model?

Jill Adams: “How it breaks a traditional model is that there’s no longer that governance. There’s the auditability, the traceability, the compliance gates, the quality gates. All of those were put in place to make sure, to your point, that in a traditional DevOps workflow, that workflow is operationalized, especially for large regulated industries, government organizations, that is just a non-negotiable for quality and trust.”

Q: What does AgentOps mean and why is DevOps not sufficient?

Jill Adams: “AgentOps, what does that mean? To us, it really means agent lifecycle management and being able to manage agents the same way that you would manage code. But that is a adoption journey that we know people are going to have to go on with us. And so some of the ways that we think of that adoption journey is first we need to make sure that organizations are ready for agents in production. And that requires some very foundational DevOps foundations like code coverage, making sure they have mature testing practices, for example, making sure they do have a governed pipeline that they can bring things to and from sandbox to production. And so really to us, DevOps starts with making sure that those agents are now in the actual workflow, in the workflow of that pipeline, so that from planning, testing, releasing, monitoring, improving, continuous improving, agents are there helping you each step of the way, and they’re managing the lifecycle of agents through that pipeline.”

Agentia’s Differentiation in a Crowded Salesforce AI Ecosystem

The Salesforce ecosystem—particularly with the introduction of Agentforce—is saturated with AI tooling. Copado’s Agentia doesn’t compete with Salesforce’s agent-building capabilities; it provides the governance layer that makes agent deployment safe, traceable, and compliant.

Q: What makes Agentia fundamentally different from other AI tools in the Salesforce ecosystem?

Jill Adams: “For us, we view the tools that Salesforce has provided not as a competition, but is a good collaboration because where we see the difference with Copado is we are now the layer that is bringing safe changes to that production org. And you can think of a system like Agentforce where you’re going to have all of these agents doing simultaneous things all at the same time. It’s kind of that agent sprawl mentality. And so you still need that governance, you still need that safety, you still need that trust. And so that’s really what Copado aims to provide is the governance, the trust, the quality gates, the auditability, so that you can adopt agents in production with the governance and guardrails that a lot of enterprises require. And we think that’s going to be just table stakes in the future as more and more agents get into production.”

ContextHub: Grounding AI in Organizational Reality

Generic AI models are trained on the vast knowledge of the internet—but they know nothing about your specific Salesforce org, metadata structure, object models, flows, or compliance policies. Copado’s ContextHub solves this by grounding agents in the actual context of your environment, dramatically reducing hallucinations and improving code accuracy.

Q: How does ContextHub ground AI agents in customer-specific environments?

Jill Adams: “We really feel like the Context Hub is one of our main differentiators that we’re bringing into this space. And so as you mentioned, with a generic AI developer tooling, it is trained on the vast knowledge of the internet, the vast knowledge in the industry. It is not trained on your specific org, your specific environment setup, your specific metadata, your flows, your object model, all of those things like the generic AI coding tools are not trained on that. So by bringing in that context and using that to ground all of our agents, it produces not just better results from less hallucinations and more accurate results, but it also makes sure that whatever it is coding is validated in your org and is able to be deployed safely. So it’s a whole new level, a whole next level, I would say, of personalization to your unique organization needs.”

Q: What specific organizational knowledge does ContextHub incorporate?

Jill Adams: “It’s not just your org, it’s also your Jira instance, your individual user stories, your deployment history. There’s so much knowledge that we’re bringing into the Context Hub, your Confluence articles, your best practice docs. So if you want your entire developer community to be coding in a certain way, you can enforce those policies via our context hub and really level the playing field and up everyone’s game. So what would be maybe like a junior developer is now coding at a capacity of a more senior developer with less hallucinations, less errors, and obviously that results in a lot more speed for the organization.”

Human Accountability in Agent-Driven Workflows

One of Copado’s core principles is that agents can execute work, but humans must remain accountable for critical decisions—especially production deployments. This human-in-the-loop model ensures that teams maintain control while still benefiting from AI acceleration.

Q: How does Agentia balance AI autonomy with human oversight?

Jill Adams: “One of our principles that we firmly believe that there has to be a human review, there has to be a human in the loop for critical processes like deploying into production or things of that nature. So one of our kind of guiding principles is that agents can do the work, but there always needs to be a human accountable at the end of the day. And so that’s part of what makes Agentia unique is that we’re able to build in those approval steps, build in those quality gates right into the workflow so that your development teams can stay in control, not lose control of what gets eventually pushed into production, but also benefit from that acceleration. And so we’re seeing a lot of great results from customers where the agent got everything maybe like 85% right. And it’s getting better all the time, the accuracy, but we still need that human to be accountable and in control.”

Q: What governance controls extend beyond code quality?

Jill Adams: “To be frank, it’s not just the code quality, it’s also let’s make sure that there’s governance around token usage. Let’s make sure there’s governance around policies and things of that nature that are not just code related. They’re best practices for your company, they’re governance on how many tokens a specific developer is using on a specific project. So it really runs the gamut from a compliance and a governance standpoint.”

Audit Trails and Accountability When AI Makes Production Decisions

When AI agents make decisions inside production delivery workflows, enterprises need clear audit trails and accountability frameworks. Copado leverages its 13-year DevOps foundation to provide the same governance scaffolding for agents that it provides for human-driven commits.

Q: How does Copado maintain audit trails when AI agents make production decisions?

Jill Adams: “I want to go back to our DevOps foundation because that is really the scaffolding on which our entire Agentia platform is built. And the scaffolding, that historical foundation of DevOps where everything does have an audit trail, quality gates are built in. That’s the same scaffolding that we’re enabling our agents now to operate within. So that’s really where we’re not reinventing the wheel here. That’s why we think we have a right to win in this space because we do have the proof that we can do this, not just for small companies, but for the largest, most complex enterprises in the Salesforce ecosystem.”

The Adoption Curve: From DevOps Readiness to Autonomous Delivery

Copado views AgentOps adoption as a journey, not a destination. Organizations must first establish foundational DevOps maturity—code coverage, testing practices, governed pipelines—before they can safely deploy agents into production. From there, the evolution moves toward continuous optimization and self-healing systems.

Q: What does the adoption journey look like for teams adopting Agentia?

Jill Adams: “The adoption curve for any sort of transformation follows a very similar model. And so one thing that we’re very cognizant of is how do we bring our customers on that adoption journey with us because everyone is going through their own adoption curve on how do they leverage AI? How do they transform into an AI native company? That’s now the imperative out there for companies. And so really what we’re doing, even from a roadmap and how we think about the future of Agentia is first we have to enable teams to get ready for agents. And so they really need that foundational DevOps foundational work just to get ready for agents.”

Q: What does the “get ready for agents” phase involve?

Jill Adams: “Agents will traverse your system. They will look into all the nooks and crannies of your metadata and they will find things that are maybe duplicative or inconsistent. And so there’s kind of this cleanup or get ready for agents phase that people need to start with. And that’s really where the DevOps foundations come in. Next, we know that customers are going to want to be able to ship agents safely into production with governance and guardrails, and that’s across the entire lifecycle. So from planning, building, testing, releasing, operating, monitoring, and continuously improving. And so that’s really going to be our focus over the next few months is getting to a place where our customers are really confident that they can ship with the right amount of agent QA, with the right amount of testing, not just the functionality, but also the behavior, because these are different systems.”

Q: How is testing different for AI agents versus traditional code?

Jill Adams: “Now they’re systems of behavior and intelligence and they’re non-deterministic. So testing is a whole different muscle with AI. So that’s really what we’re doubling down on in the next near term. And then in the future, we really see this future of continuous optimization, self-healing even in the future of these systems where we can enable teams to really supercharge their own DevOps teams. We don’t feel like this is a replacement for people. We feel like this is something that will supercharge them and keep them in control. And so that’s really our north star to get to that point where then customers can continuously monitor and improve agents in production without enormous burden on their existing teams.”