The AI Growth Playbook
A step-by-step framework for marketers and growth leaders who want to build an AI-powered growth engine, from your first prompt to a team of autonomous agents working for you.
Most people use AI like a search engine. This framework turns it into your most valuable team member.
I put this together after figuring out, the hard way, what actually works. There's a lot of AI hype out there. This isn't that. It's 6 phases, each building on the last, with the honest gaps and lessons included.
The 6-Phase Framework
Build Your Knowledge Foundation
“The quality of your AI output = the quality of context you provide. Most people skip this and blame the tool.”
Create a curated AI knowledge base using NotebookLM with Claude docs, MCP guides, Skills references, and prompt engineering best practices.
Document your business DNA: who your ICP is, your business model, competitive landscape, internal processes, terminology, and tools you use.
Document yourself: your goals, experience, thinking patterns, preferred methods, and decision-making style.
The more context AI has about your world, the less generic its outputs become. This is the step that separates average users from power users.
Your Business
How You Work
You & Your AI Setup
Start with your ICP: it's the single biggest multiplier for AI output quality.
Strategic AI Planning
“Don't use AI as a task executor. Use it as a thinking partnership. Cross-pollinate between AI systems for exponentially better strategy.”
Create a Gemini Gem loaded with your NotebookLM sources and business documentation. Give it clear instructions on the output format your team needs.
Feed the strategic output into Claude Code's planning mode for technical review. Let AI systems challenge each other's assumptions.
Research relevant Skills and MCPs before execution. These are the force multipliers that turn good projects into exceptional ones.
Specify your team's technical skill level so the AI calibrates its suggestions to what you can actually implement.
AI Cross-Pollination Workflow
Use multiple AI systems as a thinking partnership: each catches what the other misses
Strategic Planning
Plan Mode Review
Cross-Pollinate
Execute with AI as Your Co-Pilot
“Execution is where most AI adoption fails. Treat AI like a senior hire: with feedback loops, not just commands.”
Review, comment, and fine-tune AI-generated plans before executing. Never blindly accept the first output.
Iterate across AI tools: pass work between Gemini and Claude for fresh perspectives. Each system catches what the other misses.
Give constructive feedback: specifically what worked, what didn't, share your trusted sources and preferred methods.
Ask AI to learn from its mistakes and explain its reasoning. The dialogue IS the productivity multiplier.
Halfway through. The next phases are where individual productivity becomes team-level transformation.
Scale with Agents
“This is where individual productivity becomes team-level transformation. AI agents are your always-on specialized team.”
Deploy at least one agent per task or project. Each agent should have a focused domain of responsibility.
The Orchestrator Agent is your bridge: it manages the plan, tracks what's done, communicates what needs improvement, and coordinates the other agents.
A Security Agent ensures your standards are maintained across all outputs. A Research Agent continuously updates other agents with best practices.
Think of it as building a team: each agent has a role, clear deliverables, and reports to the orchestrator who reports to you.
Your AI Agent Team
Click each role to see what it does
The Human Loop
“AI amplifies your judgment, it doesn't replace it. The most successful teams build deliberate human checkpoints into their AI workflow.”
Get peers and domain experts to review your AI-assisted work. Fresh human eyes catch what AI normalizes and what you've become blind to.
Take strategic breaks. When you hit a limit or feel stuck, step back. Test what you've built. Return with fresh perspective and new ideas.
Use AI to organize your thoughts and prepare for your next execution session. The planning between sessions is where breakthrough ideas happen.
Build, build, build. Start small and go bigger. You'll be surprised how much complexity AI can handle. The limit is your ambition, not the tool.
The Work Rhythm
The most effective AI workflow isn't a marathon. It's deliberate cycles of building and reflecting
Execute with AI. Ship features, create content, run experiments.
Step back. Test what you built. Find what works and what doesn't.
Organize thoughts with AI. Prepare feedback for the next session.
Return with fresh perspective. Plan bigger. Iterate on what you learned.
Repeat → Compound → Scale
Amplify & Share
“The most powerful growth loop: build something remarkable with AI, share how you did it, attract opportunities, build something even bigger.”
Sharing is caring, and it's also the smartest growth strategy. Every workflow you document becomes content, and every piece of content builds your authority.
Your experience becomes your moat. AI tools are available to everyone, but your specific workflow and insights are uniquely yours.
Teaching others solidifies your own understanding. You'll discover gaps in your knowledge and new approaches you hadn't considered.
The compounding effect is real: the more you share, the more opportunities find you. The more you build, the more you can share.
Tal Cohen
Director of Growth · Product · AI Builder
I've spent my career at the intersection of product and growth: figuring out what to build, then how to get it in front of the right people. AI has been the biggest force multiplier I've ever worked with.
As Director of Growth at Windward (a maritime AI company that went AIM-listed and was acquired by FTV Capital). I've seen what it means to grow a fast-moving tech company. This playbook is the honest documentation of what's actually worked. No hype, no theory. Just patterns that compound.
Dad and lifelong learner. Always happy to connect, compare notes, or just chat about what you're building.
Ready to start?
Build Your AI Growth Engine
This framework works whether you're a solo marketer or leading a team of 50. Start with Phase 1, build at your own pace, and watch your capabilities compound.
Questions, pushback, or something that worked differently for you? DM me. I genuinely want to hear it.