How to build a marketing team with AI coworkers

You’re likely deep into your 2027 planning these days, so headcount, budget, and the people needed to execute your big plans for the year are likely top of mind. Beyond people, you also need to consider data and platform needs. AI-enabled roles add another consideration: How will they change team responsibilities and reporting structures? Planning for these roles today will support your 2027 plans and prepare your organization for what’s to come.

Several platform vendors have spent this year building products and features that are vying to be hired for one or more non-human roles in next year’s org chart. How many of the roles on your 2027 org chart will be non-human, and what will a truly hybrid workforce look and operate like?

Hiring your first virtual coworkers

Optimizely recently announced Virtual Teammates at their 2026 Opticon conference earlier this month. When you pull it up, the interface presents a directory you can hire from, similar to reviewing a contractor’s resume on a job site.

The roles available for hire through Optimizely’s agentic marketing platform include a chief of staff, an SEO and AI search analyst, a marketing analyst, a personalization strategist, and a CRO manager. These virtual marketers each have their own identity and permissions, can run on a schedule or on a trigger, maintain context across projects, and include an audit trail linked to that agent.

Earlier in the year, Treasure AI released a Marketing Super Agent that it describes as operating like a marketing department, orchestrating other agents inside a governed workspace. Another example comes from HubSpot, which shipped more than 20 Breeze agents and called them digital teammates, taking a shot at AI tools that behave like expensive interns. Asana’s AI Teammates come in role shapes, too, including a Campaign Strategist, with the team holding permissions and data access.

These are just four examples of companies adopting a similar architecture of identity, permissions, and an audit trail, and it’s more than a passing trend. In the past, when vendors converged like this, it meant that a category was forming. Virtual teammates are likely here to stay in 2027 and beyond.

Paying for the half you can’t buy

EY recently announced a $100 million rewards program for employees who demonstrate critical thinking skills in their work alongside AI tools. The incentive structure is a promising sign of how the human role in a hybrid future can be most effective.

Individual spot awards top out at $500. Team awards range from $10,000 to $25,000. What qualifies: judgment, business acumen, adaptability, collaboration, and experimentation with the technology itself.

Granted, $100 million is an attention-grabbing number. But what interested me more was the makeup of the incentive pool. EY valued team play at 20 to 50 times the individual prize.

With this huge financial stake in AI-enabled output, EY recognizes that highly paid human capability is about shepherding a human-machine hybrid to deliver something its clients will accept. Similarly, KPMG is reimagining its audit intern program with an emphasis on critical thinking. PwC is teaching its people AI proficiency alongside empathy and innovation. Within 12 months, three prominent consultancies have redefined what they reward and clarified expectations for humans in a hybrid human-AI work environment.

Companies looking to implement something similar should assign clear accountability, with identifiable points of responsibility for the responsible stewardship of AI-generated output.

A prediction for next year

Gartner estimates that over 40% of agentic AI projects will be canceled by the end of next year due to escalating costs, unclear ROI, and a lack of governance over risk.

That same Gartner report estimates that, out of the thousands of vendors selling agentic AI, roughly 130 are actually building truly agentic capabilities into their platforms. The rest are applying the “agentic” label to rebranded assistants, chatbots, and RPA workflows without truly changing how they function.

That means marketers investing in so-called agentic platforms and tools may not be getting what they’re paying for, further contributing to disillusionment with agentic AI projects that fail to yield the anticipated results.

That might sound like a caution against over-indexing on the idea of planning teams around agentic teammates. Another stat on that Gartner slide deserves attention: the research firm also forecasts that 33% of enterprise software applications will include agentic capabilities by 2028 — up from less than 1% in 2024 — and that at least 15% of day-to-day work decisions will be made without human intervention. 

Consider that, at about one routine decision for every seven made, that 15% figure represents a solvable design challenge. With the right planning around agents, marketing leaders can keep agency in their own hands while deploying bots to handle the dull stuff.

Deciding what to delegate before deciding what to take control of

On the surface, a company investing in an agentic platform that promises AI teammates might seem like just any other software purchase, but it’s not. Few companies have a strong track record of planning the people and process side of platform acquisition. This can lead to procuring a martech platform with a ton of features and unclear responsibilities for who manages what.

Adding virtual team members makes this even more problematic, so the pre-purchase phase should resemble how you define job roles before hiring a human team member.

A good place to start is thinking about how you’ll manage each tool or task. Path-based management means you specify the 20 steps of a nurture sequence. Goal-based management means you set the intent and the guardrails, e.g., lift mid-market SQL conversion 15%, hold brand sentiment at 4.0 or better, never touch an account with an open Tier 1 ticket, and let the agent find the route inside them. Every product above is very good at finding routes, but none can (or should) do the planning.

When you’ve completed your thinking on the management side of the task, assess the work and decide what you can delegate. Ask yourself three questions about each potential task to identify which tasks are agent-ready and which require human oversight:

• Is it reversible? Low-risk work you can undo tolerates real autonomy. Anything irreversible or trust-sensitive should be under review.
• How is output reviewed before customer contact? Rather than leaving the answer as “nobody” due to an oversight, be intentional about this based on risk.
• What happens when two agents disagree? A pricing agent and a retention agent will eventually chase competing objectives on the same account, and someone has to decide in advance who wins.

Let’s revisit the Optimizely Virtual Teammate org chart for a minute. An AI Marketing Analyst converting website or campaign data into insights that yield easy decisions ranks pretty high on the “Autonomy” axis. But a Personalization Strategist who codes, tests, and deploys live campaigns deserves several rungs higher and some scrutiny: same folder. Same afternoon. Entirely different levels of autonomy.

Scaling this approach to adopting agentic teammates also has the benefit of requiring minimal bureaucracy, which is its own reward. One human leader can be responsible for a small number of specialized agents grouped by outcome rather than channel.

That leader has one very human job: validating agent logic for drift, defining context that agents can inherit from one another, arbitrating conflicts, integrating new agents into a sandbox before giving them live access, and hitting the kill switch if needed. Title that role, add it to the org chart.

Planning a sequence before optimizing for speed

Thinking through the process ahead of time and taking a methodical approach will pay off. Think of this as a three-step process: stabilize, standardize, then automate. If you’re still debating what a qualified lead looks like when you’re in the middle of implementation, you’ll pay for the gain in velocity with downstream inefficiency (not exactly tech debt) when you add agents into the mix, because velocity on a fragmented operating model mostly means faster entropy.

You can hire a virtual teammate in an afternoon, yet engineering your org to absorb that one new teammate takes a quarter or two, which is why hiring an agentic teammate still takes careful planning to metabolize into your team and its processes.

This is where a return to the fundamentals of measurement and experimentation pays off. Figure out how you’ll measure it before kickoff, instead of assuming it can be figured out as you go.

Does a managed ecosystem of humans + bots produce enough incremental value to justify its cost, its complexity, and the governance it requires? Agent utilization and the percentage of campaigns completed can be distractions or, at best, vanity metrics. In fact, vague business value was a top-cited reason for cancellation in Gartner’s study, and in my opinion, that describes a measurement and strategic planning issue far more often than a technology shortcoming.

How this should affect your 2027 planning

Even if you consider a new type of hire for your team, that doesn’t mean humans and agents are interchangeable. Still, it’s increasingly useful to step back and realistically assess the roles and tasks you need on your team, and whether a human alone or a human supported by an agent (or team of agents) is the best solution.

This doesn’t change humans’ strategic and operational roles. An agent can tell you what happened and what’s likely to happen next. Deciding what your organization should do about belongs to someone who can be held accountable. 

Hand that off to a directory listing, and you’ve outsourced strategy while believing you bought capacity. The platforms are the easier part here, as they usually are in any substantive change. People and processes determine whether this succeeds in the long term.

This naturally leads back to the question of how many boxes on your 2027 org chart will be non-human and who (or what) will manage them and help improve them?

Team planning will need to be more of a hybrid exercise than ever before. Begin by identifying roles on your human team that will manage AI agents. Then, evaluate every candidate task for autonomy fit before completing any platform purchases.

Budgeting in a binary human-or-software way may not work moving forward, and success will mean collaboration between humans and agents. This will complicate your 2027 planning, but is only the beginning of this new way of team planning.

The post How to build a marketing team with AI coworkers appeared first on MarTech.