Key takeaways
- B2B marketers are rapidly becoming “citizen developers,” building AI workflows that automate research, personalization and handoffs without waiting on engineering teams.
- Responsible AI adoption depends on secure environments, runtime visibility and clear operating boundaries so experimentation can scale without creating data, brand or performance risk.
Your marketing team is building software. Maybe you haven’t called it that yet, but that’s what’s happening. Across B2B organizations, marketers are wiring together campaign automations, connecting customer data to content systems and shipping AI-powered workflows that touch real revenue. The pressure to transform with AI is relentless, and sitting on the sidelines isn’t an option anymore.
Research shows 67% of employees want their organizations to use more AI. That enthusiasm is real. But 36% still don’t understand why they’re expected to use it, and that gap is where value quietly slips away.
A new kind of builder has emerged from this shift: the citizen developer. These are marketers with no engineering training who now construct agents, workflows and applications that once required a full technical team. A demand gen manager can build an AI workflow that researches every event attendee, cross-references CRM history, identifies your highest-potential prospects and routes personalized outreach to the right salesperson. Powerful stuff. The problem starts when an experiment becomes operational software without anyone treating it that way, complete with credentials, customer data and permission to act across your systems. Move fast, and you risk data security, brand consistency and campaign performance breaking all at once. Slow everything down with approval tickets, and your best builders stop building. The teams that win will resolve that tension, not ignore it.
To unpack how marketing leaders can navigate this moment, we sat down with BlueRock CMO David Greenberg. He’s spent his career at the intersection of marketing transformation and responsible technology adoption, and he doesn’t deal in slogans. In this conversation, Greenberg breaks down three subjects every CMO should be thinking about right now: the rise of the citizen developer and what these marketers actually build week to week, what responsible AI adoption looks like in practice versus the guardrails-as-buzzword trap most teams fall into, and the concrete first moves for turning casual prompting into secure, repeatable AI workflows that scale.
Demand Gen Report (DGR): Research shows that 67% of employees want their organizations to use more AI, yet 36% still don’t understand why they’re expected to use it in their roles. For a marketing leader, which number is the bigger red flag, and why?
David Greenberg: The 36% is the bigger red flag. Enthusiasm for AI is important, but adoption without a clear understanding of the problem you are trying to solve creates a lot of activity without necessarily creating value.
Marketing leaders should be helping their teams identify where AI can materially change how work gets re-imagined with these new tools.. That might mean automating a repetitive research process, building a workflow that identifies and enriches prospects, or creating a system that continuously analyzes campaign performance. Once employees understand the problem they are solving, AI stops being another tool they have been told to use and becomes a way to build something useful.
The goal should not be simply to get more marketers using AI. It should be to help them recognize where they can use AI to redesign work, remove friction and build new capabilities that did not exist before.
DGR: You use the term “citizen developer” to describe marketers building AI workflows without engineering training. Walk us through a specific example. What does a citizen developer on a demand gen team actually build in a given week, and where does it quietly go wrong?
Greenberg: A demand gen manager might build an AI workflow around event follow-up. It could take an attendee list, research each company, combine that with CRM history and engagement data, identify the highest-potential prospects, draft personalized outreach and route those insights to the right salesperson. What used to require several tools and a lot of manual work can increasingly be built by the marketer who understands the problem best.
That’s the bigger shift. Marketers no longer have to accept the limitations of the software they buy. They can increasingly build the applications, agents and workflows they need themselves.
Where it quietly goes wrong is when an experiment becomes operational software without anyone treating it that way. That workflow may now have credentials, customer data and permission to take actions across multiple systems. Organizations need to enable these new citizen developers, but also give them safe environments, practices and guardrails so they can experiment and move into production without creating unnecessary risk.
When Do Marketing Teams Cross From Using AI to Building Software?
DGR: Marketing teams have quietly become software builders, wiring together campaign automations, connecting customer data to content systems and shipping AI-powered workflows that touch real revenue. Do most CMOs realize their teams have crossed that line, and what happens when they don’t?
Greenberg: CMOs are under enormous pressure to transform marketing with AI. Sitting on the sidelines isn’t really an option anymore. But moving fast and transforming well are two different things.
Many leaders recognize their teams are using AI, but fewer recognize they are increasingly building software. Using AI to create content or analyze data is fundamentally different from building agents and workflows that connect systems, make decisions and take actions.
That’s the new challenge for the CMO. They need to push their organizations to reimagine how marketing works, while making sure teams have the skills, safe environments and operating practices to build responsibly. The mandate isn’t simply to adopt AI faster. It’s to transform the function without creating a new set of problems along the way.
DGR: When marketers adopt tools faster than leadership can govern them, what specifically breaks first: data security, brand consistency or campaign performance?
Greenberg: It can be all three, and that’s what makes this different. With traditional marketing technology, those risks were often separate. With AI systems that can make decisions and take actions, one unexpected change can affect customer data, alter the brand experience and impact campaign performance at the same time.
The real problem is that teams may not know something has changed until the outcome is already visible. Agents can adapt as they execute, so what worked yesterday may behave differently tomorrow.
That’s why teams need both a secure workspace for AI to build and operate within defined boundaries, and rapid visibility into what those systems are actually doing. You want to give AI room to work, while being able to quickly understand when behavior changes and correct course before a small issue becomes a much bigger one.
What the AI Supply Chain Means for How Marketing Teams Build
DGR: Recent incidents involving OpenAI models and Hugging Face exposed real vulnerabilities in the AI supply chain. Why should those headlines change how their team builds this quarter?
Greenberg: The lesson isn’t that marketing teams should stop using AI or become experts in AI security. It’s that they should stop assuming an AI tool is a self-contained application. Modern AI systems increasingly depend on models, APIs, packages, connectors, MCP servers and other external services. Every connection expands what the system can do, but it also expands the environment the organization needs to understand.
Teams need a secure workspace where AI can build and operate within defined boundaries, with runtime controls and real-time visibility into what it’s doing and when behavior changes.
AI is moving incredibly quickly. Trying to evaluate every possible risk before allowing people to build won’t scale. Creating sensible boundaries around execution will.
DGR: “Responsible AI adoption” gets used as a slogan far more than it gets defined. What are three things a marketing team doing this well has in place that a team doing it poorly doesn’t?
Greenberg: First, they give people a secure place to build and experiment. Teams need room to try things, make mistakes and learn without unnecessarily exposing sensitive data or production systems.
Second, they have real visibility. They know what AI systems are running, what they can access and what they’re actually doing as they execute.
Third, they establish clear boundaries without creating an approval process for every new idea. Builders know where they have freedom to move quickly and where additional controls are needed.
Responsible AI isn’t about slowing people down. It’s about creating an environment where teams can build faster because the right foundations are already in place.
DGR: Guardrails often feel like the enemy of speed, and speed is the whole point of AI for marketers. How do you build guardrails into campaign and automation workflows without turning your best builders into people who wait on approvals?
Greenberg: The mistake is treating governance as an approval process rather than an operating model.
If every new workflow requires a security ticket and three meetings, people will either stop building or find ways around the process. Neither outcome is particularly useful.
A better approach is to establish boundaries around what systems and data AI can access and what actions it can take, then give builders freedom inside those boundaries. Low-risk experimentation can move quickly. Higher-impact actions can receive additional controls.
The goal is to move governance closer to the point where AI actually takes action rather than forcing humans to approve every step beforehand. That’s especially important as agents become more autonomous.
Where AI Workflows Break When You Connect Multiple Systems
DGR: Marketing AI workflows increasingly pull from CRMs, product data and customer engagement platforms all at once. Where’s the most overlooked exposure point when a marketer links those systems together, and how would you close it?
Greenberg: The most overlooked exposure is often the connection between systems. The CRM may be trusted, the AI model may be trusted and the marketing platform may be trusted, but connecting them can give an AI workflow far more access than anyone intended.
That’s why teams need to limit permissions, maintain visibility into what the AI is doing, and run these workflows in a controlled workspace that can enforce boundaries as they execute. The key question isn’t just, “Do we trust each application?” It’s, “What can the AI do once they’re all connected?”
DGR: Suppose a CMO wants to move their team from casual prompting to secure, repeatable building. What are the first three moves they should make in the next 30 days, and which one do most leaders skip?
Greenberg: First, identify a handful of real workflows worth building. Look for repetitive work, manual handoffs or processes where employees already understand the problem deeply.
Second, give teams a secure environment designed for agentic building, where they can experiment and run AI with appropriate boundaries, visibility and controls. This is the step most organizations skip, even though these environments are increasingly accessible.
Third, teach people how to build well. That means hands-on learning around AI development best practices, not simply better prompting.
The goal is to combine real business problems, a safe place to build and the skills to turn experiments into repeatable systems.
DGR: What separates the teams that turn AI into a durable competitive advantage from the ones that end up cleaning up a breach or a compliance mess?
Greenberg: The strongest organizations will recognize two shifts: agentic AI requires a different operating model, and citizen developers are becoming a real part of how the business builds.
The companies that get this right will actively enable those builders across marketing, sales, finance and operations, giving them safe environments, visibility and clear boundaries to move quickly.
The long-term advantage will come from embracing citizen developers while putting the agentic operating practices in place that allow them to build responsibly and at scale.
Leave a Reply