פרק 32: מה חייבים לדעת לפני שבונים מוח מרקטיאלי

Summary

The episode explains how to build a “marketing brain”: a shared, continuously maintained system of company knowledge, context, and workflows that helps teams produce consistent work at greater scale with AI.

The goal is not simply to create another knowledge base or collection of AI tools. It is to turn expertise into operating infrastructure—like a restaurant’s recipes, stations, standards, and processes—so quality does not depend entirely on who happens to do each task.

Key Ideas and Playbooks

• Treat context as the foundation. Gather and organize what AI needs to know: product and company information, positioning, messaging, customer and market insights, competitors, examples, and standards for good work. Structure this into relevant building blocks instead of giving every task an undifferentiated dump of information.
• Build a living system, not a one-off project. The brain must be updated as products, markets, competitors, and organizational practices change. There is no finished state; teams need ongoing iteration, human judgment, and quality control.
• Move from one-off outputs to repeatable “factories.” Identify recurring tasks and create workflows that produce useful outputs repeatedly. The aim is not just to generate more content, but to reduce manual work and make reliable work available at scale.
  • Internal factories help marketing teams research, synthesize information, and do their own work.
  • Self-service factories let Sales, Customer Success, or other teams generate materials, such as one-pagers, without waiting for Marketing, with appropriate review and training.
• Use a three-part operating model:
  1. Builders manage the technical infrastructure, integrations, automations, and skills. Not everyone needs to be a builder.
  2. Context owners and producers supply and maintain the information, methods, examples, and standards the system needs.
  3. Operators and reviewers assess outputs, catch errors or low-quality work, route specialized items to the right experts, and feed lessons back into the context.
• Create a feedback loop. People with relevant expertise should review AI-generated work. Their corrections should improve the underlying context and workflows, making future outputs more accurate and useful.
• Pair enablement with implementation. Training alone may not translate into team-level impact, while deploying tools without training and maintained context can lead to stale or poor outputs. The guest recommends combining practical training with systems teams can use in their real work.

Examples Discussed

• Competitor battlecards: Monitor competitors’ releases and changes, summarize relevant signals, and alert the team when something important happens. Potential sources include competitor websites and help centers, as well as internal sales conversations and customer feedback.
• Product-release summaries: Feed release information into a workflow that extracts relevant points for Marketing and Sales, reducing manual follow-up with Product.
• Sales and customer insights: Analyze recurring questions and objections from calls, emails, and other touchpoints, then make useful answers available to the wider team.
• Self-service sales materials: Let Sales or regional marketing teams create a first draft of a one-pager or event messaging from shared context, with a human reviewer checking quality.
• Previously neglected projects: A well-contextualized system can make large tasks—such as building or refreshing a website glossary—feasible in days rather than leaving them as long-running projects. This was presented as an example, not a measured performance guarantee.
• New customer-facing formats: With repetitive production handled more efficiently, PMMs may have more room to experiment with interactive or tailored materials rather than relying only on familiar formats such as one-pagers.

Actionable Recommendations

• Start with an existing shared folder or drive. A basic set of well-maintained context files is a useful first version.
• Make files modular and task-relevant—for example, separate company context from competitor information—to avoid irrelevant context leaking into outputs.
• Connect useful sources such as recorded calls, email, CRM data, and product information where appropriate. Capture important moments or action items so they can be found later.
• Define what good and bad output looks like. Provide concrete examples and explain terminology, tone, priorities, and nuances; do not assume an AI system understands a request just because it sounds clear to a person.
• Begin with tasks the team already performs. Test the workflow internally, review results, and refine it before making it available for self-service.
• Schedule a regular review of incoming insights and updates, and decide what belongs in shared organizational context.
• Route sensitive or technically complex outputs to qualified approvers. Reviewers should also identify when the underlying context—not just the generated output—needs correction.
• Focus on business impact rather than the novelty of building an agent or skill: Which work becomes faster, more reliable, or possible that the team could not previously prioritize?

Metrics, Targets, and Timelines

• The guest said the 10X Marketers community has more than 2,000 marketing managers. The host described the PMMiya community as having more than 6,000 members.
• A suggested operating cadence was about 30 minutes per week to review and refine updates. Competitor monitoring and summaries were also discussed as potentially running weekly, with alerts for significant changes.
• A practical workshop offered by the guest’s company consisted of five sessions. The discussion says teams were revisited after several weeks or months, but gives no measured adoption or business-impact figures.
• A 10X conference on November 26 was mentioned; no year was specified in the subtitles.
• No revenue, CAC, LTV, churn, or quantified productivity targets were provided. The implied measures of success are output quality and consistency, adoption, freshness of context, reduced manual effort, and the ability to address more work—not numerical targets stated by the speakers.

Presenters and Sources

• Orian Aviner, host and founding partner of the PMMiya community.
• Noy Rubinstein, guest and co-founder of 10X Marketers, alongside Adar Yakir and Alex Grinshpon.
• The discussion draws on the speakers’ work with marketing teams and organizations building AI-supported marketing operations.