The pressure to deliver higher results with leaner teams is a constant obstacle in modern marketing operations. While AI agents can handle research, reporting, and first drafts, delegating the strategic context behind that work is much more complex. An agent can’t recover why your brand avoids certain language or remember a sensitive customer story unless your team captures it.
Managing this change was the focus of the article “Meet your new teammate: Building marketing workflows around AI agents” at the November 2026 MarTech conference. Industry leaders explored how SEO, MOP, and marketing leaders can share workloads with AI while protecting the human judgment that drives real business value.
Stacey Ackerman, independent consultant and fractional CMO, moderated the discussion with Ed Poppe, founder and leader of fractional marketing at Poppe Marketing; Sean O’Connell, head of customer success at Screendragon; and Lisa Shaw, founder of Devon Point Group.
A survey of the public highlighted how widespread this transition is: 51% of respondents reported that their organizations are actively experimenting with AI, and none reported avoiding it. Here is the path to follow to make these pilot projects sustainable.
Document the “why” behind your marketing decisions
When headcount shrinks and departed team members are not replaced, remaining staff face intense delivery expectations. Beyond the immediate workload, this creates a quieter risk: Institutional knowledge goes away.
Shaw uses AI agents to coordinate activities and compile reporting data, freeing its attention for high-value strategic work.
“What the AI is doing is freeing up my attention so I can spend it on judgment and relationships,” he explained.
To keep its agents aligned, Shaw documents the reasoning behind past decisions, capturing brand nuances, previous customer friction, and historical context, and inserts these lessons directly into the AI’s instructions. “I think the why is as important as the what,” he noted.
Poppe linked this practice to a key marketing strength you already have: writing a strong creative brief. Defining deliverables, articulating brand guardrails, and marking checkpoints for review remain essential when briefing AI. Your strategy documentation is what turns a generic tool into a reliable teammate.
Fix the workflow before automating the task
Accelerating an isolated task can inadvertently push bottlenecks downstream to other teams. O’Connell highlighted the difference between the overall workflow (the end-to-end work path) and the individual agents that handle repeatable steps within it.
In organizations ranging from single experimenters to enterprise teams with agents in production, the biggest challenge is connecting disconnected pilots into a cohesive system. Rather than starting with a genius tool, O’Connell recommends first identifying the operational bottleneck and defining the characteristics of success. “Start with the process, not the technology,” he said.
Ackerman reinforced this approach: Overlaying AI on top of a broken manual process only amplifies friction. As you build better workflows, Poppe encourages the involvement of cross-functional partners. Sharing a working pilot gives hesitant colleagues on sales, operations, or creative teams a concrete model to learn from, turning isolated experiments into shared victories.
Keep customer-facing judgment in human hands
Drawing a clear line between agent execution and human ownership protects both your brand reputation and customer relationships. Shaw offered a rule of thumb for delegation: “The closer a task gets to the customer, the more it needs a human in the loop.”
Agents excel at extracting data, formatting assets, and assembling initial drafts. However, executive communications, sensitive renewals, and messages that require genuine empathy belong to people. O’Connell agrees that final approvals should rest with human leaders: agents can accelerate the preparation work that leads to a decision, not own the outcome.
Even well-trained agents need constant guidance, as production can deviate from brand over time. Shaw likens AI to an overconfident intern whose polished-looking drafts still require scrutiny. Incorporating a critical editorial review ensures that your content remains truly relevant and persuasive to your audience.
Integrate quality controls and governance into your system
Sustaining AI performance requires continuous calibration and this is achievable with lightweight guardrails. O’Connell explained how teams can implement a rating layer, where a secondary AI model scores agent outputs against established brand criteria before flagging items for human review. Tracking performance over time also helps you spot when changing models might improve accuracy or reduce costs.
Governance also means collaborating with your team rather than blocking tools. During the audience Q&A, O’Connell noted that strict bans rarely prevent employees from using personal AI accounts to meet deadlines. Providing secure, approved access allows your team to innovate safely instead of working in the shadows.
Building resilient AI workflows comes down to focused experimentation and open collaboration: pick a real bottleneck, define clear boundaries for your agents, and share what works in your organization. As Poppe summed it up, “So learn and share.”
