What changes when AI can run inside your Marketo instance

This article summarizes the most important insights, examples, and takeaways from Kapturall’s webinar, “What Changes When AI Can Run Inside Your Marketo Instance.”

During the session, we explored how AI agents and LLMs are beginning to operate directly inside Marketo Engage — transforming operational workflows, governance processes, campaign execution, and the skills modern marketing teams will need moving forward.

The advent of AI is fundamentally shifting the skill sets required for successful Marketo practitioners.

Traditional strengths like UI speed and rote execution of tasks such as cloning or basic translation, are diminishing in value. The focus is moving towards higher-order thinking and strategic capabilities. 

Skills gaining importance include the ability to frame the right questions, understand the logic behind operational decisions, and exercise critical judgment to evaluate AI-generated outputs.

A deep understanding of the Marketo data model is crucial, as it underpins the ability to provide precise instructions to AI and recognize incorrect results. Continuous AI literacy is also essential. 

“What is gaining value? Asking the right questions in the right way. If I need to choose one thing, that is it.” J. Tarzian 

Key takeaways: 

• Shift focus from UI speed and routine tasks to strategic thinking. 

• Develop skills in asking precise questions and framing operational logic for AI. 

• Cultivate strong quality judgment to identify and correct erroneous AI outputs. 

• Deepen your understanding of the Marketo data model to guide AI effectively. 

• Prioritize continuous AI literacy, dedicate time daily to learning new advancements. 

Jose demonstrated how an LLM can translate an existing English email into Spanish and automatically create a new, fully translated email within the Marketo UI.

The process involves providing the AI with the original email ID, the desired new email name, and the target language. 

The demonstration highlighted the efficiency of this approach for global marketing teams, as it can be scaled to multiple languages.

Notably, the AI displayed an understanding of what content should not be translated, such as brand names or marketing jargon, ensuring accuracy and brand consistency. 

 “The model also understands what should not be translated — for example marketing jargon or brand names.” J. Tarzian 

Key takeaways: 

• Leverage AI to automate the translation of existing Marketo emails into multiple languages. 

• Use specific prompts including email ID and target language to guide the translation process. 

• AI models create new, translated emails directly within the Marketo UI. 

• Advanced AI preserves non-translatable content like brand names, maintaining brand integrity. 

• Streamline email localization workflows, reducing manual effort and potential for errors. 

Jose demonstrated a practical application of AI for Marketo reporting, specifically analyzing email template usage.

By leveraging an LLM, we are able to quickly identify the most frequently used email templates within a specified timeframe, providing insights into their operational impact. 

The key was not simply asking the AI a question, but guiding it with detailed, step-by-step instructions. This significantly reduced API calls consumed and expedited the analysis.

The output revealed that a substantial percentage of emails were based on a single template. 

“What I learned is that you need to guide the steps. The output in this case: 151 emails using this template, 89% of emails are based on this one template.” 

Key takeaways: 

• Use AI to analyze Marketo asset usage and identify the most-used ones. 

• Craft detailed, step-by-step prompts for greater efficiency and fewer API calls. 

• Gain quick insights into the impact of template changes by understanding usage patterns. 

• Leverage AI to inform asset management and content strategy decisions. 

While AI offers significant potential for Marketo operations, practitioners must be aware of its limitations.

Challenges extend beyond API coverage to data quality and governance. Poor data quality or a lack of clear naming conventions will lead to increased API calls, more errors, and less fluid AI performance. 

A significant risk is the expectation of “magic” from AI. Simply providing a brief prompt will not yield a fully functional outcome.

A deep understanding of the Marketo data model and clear objectives are paramount, practitioners need to understand how Marketo works under the hood to provide proper instructions and accurately judge AI outputs. 

 “The biggest risk I see is expecting magic to happen. You go in, put a two-line prompt, and expect a full engagement program — successful, engaging, and beautiful. That is not going to happen.” J. Tarzian 

Key takeaways: 

• AI limitations in Marketo extend beyond API coverage to data quality and governance. 

• Ensure robust data quality and clear naming conventions to optimize AI performance. 

• Avoid expecting “magic” — clear objectives and detailed prompts are essential. 

• Cultivate a deep understanding of the Marketo data model to instruct AI and validate outputs. 

• Don’t let organizational AI policies block initial exploration — start somewhere. 

AI-powered governance agents can be configured to check and fix compliance-related elements in Marketo emails, such as footers.

This enables automated scanning of newly drafted emails to ensure adherence to data compliance laws like CAN-SPAM or GDPR, specifically verifying the presence of elements like privacy policy links. 

We described how an agent could flag non-compliant emails, notify the user, and potentially perform automated corrections before the email is approved for sending.

The feasibility depends less on the AI server itself and more on the underlying process architecture and the Marketo API. 

 “You can even ask it to fix it before approval. It is not that simple but it is doable.” 

Key takeaways: 

• Configure AI governance agents to check email footers for CAN-SPAM or GDPR compliance. 

• Leverage AI to scan new emails for missing elements, such as privacy policy links. 

• Automate notifications when emails do not meet compliance requirements. 

• Explore AI agents automatically fixing compliance issues before approval. 

• Design process architecture around Marketo API capabilities for effective AI-driven compliance. 

Jose demonstrated a refined approach to using AI for Marketo governance, targeting programs missing essential tags.

The challenge involved identifying programs created within a specific timeframe and folder that lacked required country or solution area tags. 

Through an iterative process, providing the AI with a highly detailed, step-by-step prompt significantly improved efficiency and reduced API calls.

This granular guidance allowed the AI to execute the task precisely, surfacing the programs needing attention and enabling automated notification to the responsible team member. 

 “Now I know which programs are missing the required tags. I can go to the person in charge of that folder.” 

Key takeaways: 

• Use AI to identify Marketo programs missing required tags for better governance. 

• Craft detailed, step-by-step prompts to enhance efficiency and reduce API consumption. 

• Optimize prompt engineering for complex operational audit tasks. 

• Direct AI to identify untagged programs within a defined folder and timeframe. 

• Automate the resulting notification to the person responsible for that partition. 

When integrating LLMs with Marketo, practitioners should prioritize paid subscriptions for enhanced data privacy and security.

While free versions exist, paid tiers offer better protection for sensitive company data and more robust functionality, essential for enterprise operations. 

It is also critical to track API calls consumed by any LLM integration. An LLM server connects via the Marketo API and contributes to the daily limit of 50,000 calls.

Neglecting this can lead to exceeding daily limits and disrupting other vital Marketo integrations and external reporting systems. 

 “One thing I would always say: you need to keep track of API calls consumed by the MCP session.” 

Key takeaways: 

• Use paid LLM subscriptions for improved data privacy, security, and functionality. 

• Address company-specific security and privacy policies before integrating LLMs. 

• Track API calls consumed by LLM integrations to avoid exceeding daily limits. 

• LLM server connections consume Marketo API calls — factor this into your daily quota planning. 

• Manage API consumption to prevent disruptions to other critical integrations. 

You can still watch the webinar on demand and see exactly how AI agents are beginning to operate inside Marketo Engage, from multilingual email creation and governance automation to smarter operational workflows.

Whether you are evaluating AI for your team or already experimenting with it, this session will help you understand what is practical today, what requires caution, and which skills will matter most moving forward.

Watch now the webinar On-demand.