AI native marketing | Live workshop: Marketing Brain, context engineering
Generic AI output is a context engineering problem.
A GitHub repository context library: your ICP, positioning, voice and proof, wired into Claude Code and n8n via MCP, which connects AI to your tools.
- Work out which context a model needs and which is filler
- Structure files so AI finds the right one without being told
- Test every change so you can prove it moved the output
- Extract brand voice from work you have published, not from a brand deck
- Write quality criteria the model applies to every draft
Replaces: Pasting context into every prompt, and the generic drafts that come back when you don’t.
Workshop dates:
Tuesday, 29 September and Tuesday, 6 October 2026
4 PM CEST / 9 AM CDT / 2 PM UTC | 90 minutes each
Trusted by marketers at
Meet your instructors:
4x’d efficiency & leads CXL’s growth, experimentation & product. Trained 25k+ marketers with 200+ courses, 4000+ blogs & playbooks.
Content Lead at CXL, focused on practical applications of AI in marketing. Trained 2000+ marketers from Unilever, Red Bull, Heineken and more.
The cause is generic context. Many marketers still start every session with a blank page.
In CXL’s panel of 80 B2B marketers, 42% named building agents as the skill they want most, the strongest single signal in the survey. The same people rate themselves lowest: 65% say they’re beginners or have no skills yet.
The gap doesn’t close with prompting, but with architecture. When the model knows your ICP, your positioning, your voice, and the proof behind your claims, the first draft lands close, and every workflow you build on top of it inherits that grounding.
You leave with the library live, connected to your tools, and a before-and-after test that shows what changed. The five weeks that follow all read from it.
📅 Two sessions, 90 minutes each
Session 1: Tuesday 29 September 2026
Session 2: Tuesday 6 October 2026
4 PM CEST / 9 AM CDT / 2 PM UTC
Part 01: What context a model needs
Before you build anything, you see which context genuinely moves an output and which is filler that costs tokens and changes nothing.
- The five context types that change output quality: ICP, positioning, voice, proof and constraints
- Why a style guide written for humans fails as model context, and what to write instead
- Outcome: A written inventory of the context you have, the context you need, and the gap
Part 02: Structure files so AI finds them
A context library the model cannot navigate is one the model ignores. You build the folder structure and the constitution file that tells it what to load before you say a word.
- The folder structure: ICP, positioning, voice, proof, and the constitution file that indexes them
- Naming and file-size rules that keep retrieval accurate as the library grows
- Outcome: A structured GitHub repository context library with your first files in place
Part 03: Wire the library into Claude Projects, Claude Code and n8n
A context system should work across all your chats. Connect it to the tools you already use so it loads automatically.
- Wire the library into Claude Projects and Claude Code, and confirm it loads on every session
- Connect three of your own tools through MCP, so workflows read the same source
- Outcome: Your context system running live in Claude Projects, Claude Code and n8n
Part 04: Encode voice and quality criteria
Voice extracted from your published work beats voice described in adjectives. You build the files that hold your standards and apply them on every output.
- Extract voice patterns from work you have already published, not from a brand deck
- Write quality criteria the model can apply, with examples of pass and fail
- Outcome: Voice and quality files the system applies to every draft
Part 05: Prove the context changed the output
You finish by testing it. Same brief, same model, with and without your context, scored against criteria you set.
- Build a simple before-and-after test you can rerun whenever the library changes
- Outcome: Evidence the system works, and a test you rerun every time you add to it
What you need before you start
- You should be comfortable with: running campaigns and reading your own performance data. No coding needed.
- You do not need: prior Claude Code, n8n or GitHub experience. Every build starts from a template.
- A paid LLM account, Claude Code preferred, plus free GitHub and Miro accounts.
- The Monday prep in the app, matched to your level by the quiz.
- Any other tool is explained live, with alternatives. You are not locked into one stack.
What marketers say after our workshops and courses
AI-Forward Marketing Executive
“I was in your workshop 12 hours ago. I already have a workflow live… I feel like a freaking WIZARD!”
Growth Marketer | MBA | Product Owner
“What a great session, and what an incredible week at CXL… what I enjoy the most is this:
Being around curious minds, leaving inspired to test, putting ideas into practice. having conversations that sharpen your thinking, asking (a lot of) questions, sorry, not sorry…
The SEO + n8n session gave me a lot to think about… honestly, I feel like a machine.”
AI/B2B Marketing Programs
“Officially CXL Certified in Redesigning Marketing Workflows for AI.
A great deep-dive into building AI into marketing the right way… rethinking the workflows from the ground up.”
PR & Marketing | MBA
“How is AI affecting the role of a marketer? I have seen this question asked a hundred times and rarely seen a credible answer. CXL ran a webinar this week that actually delivered one.
It had a bit of a cold shower effect, just data and a clear-eyed look at where things are.”
Growth Experimentation | 500+ A/B Tests
“CXL changed my life. I’m not exaggerating, it actually did…
I got so deep in the content that a year later, in a completely unexpected turn of events, I was hired by Speero. Boom. Easily a 50x ROI right there.
CXL doesn’t mess around. They prioritize quality over quantity to make a real impact on your career.”
Take this 2 session workshop on its own, or join the 6 week cohort.
It runs standalone, and as weeks 1 and 2 of the AI Native Marketer cohort: five working workflows across six weeks. Cohort enrollment closes before the first workshop.
Join our next AI native marketer cohort
Six weeks. Five working workflows. Starts 28 September 2026.
4PM CEST / 9AM CDT / 2PM UTC
Marketing Brain
You buildA context system on a GitHub repository: your ICP, positioning, voice and proof, wired into Claude Code and n8n via MCP, which connects AI to your tools.
It replacesPasting context into every prompt, and the generic drafts that come back when you don’t.
4PM CEST / 9AM CDT / 2PM UTC
Campaign engine with a quality gate
You buildAn automated workflow: one brief in, a full campaign out (emails, social, enablement, sequences) with sources and a brand voice check before it ships.
It replacesThe generic first draft, and the reviewer who catches what AI made up.
4PM CEST / 9AM CDT / 2PM UTC
AI visibility and AI-era measurement
You buildA share-of-model baseline, how often ChatGPT, Gemini and Perplexity name you, plus AI traffic grouped in GA4 and a citation-gap list of pages to write.
It replacesThe SEO specialist’s monthly report, which misses where buyers now shortlist.
3PM CET / 9AM CDT / 2PM UTC
Marketing insight and reporting agent
You buildAn agent that reads your ad platforms, GA4, CRM and email, then posts one weekly summary: what changed, why it matters, what to do about it.
It replacesThe weekly report you build by hand, and the analyst who explains it.
4PM CET / 9AM CST / 3PM UTC
Company Marketing Brain
You buildThe same context system customized for your team: one versioned source every AI tool draws from, with permissions and a rollout plan.
It replacesEveryone’s private prompts, and the colleague who knows where the answer lives.
Build one AI native workflow every week that you use at work.
You build a context brain your AI draws from, a campaign engine with a quality gate, an AI search visibility baseline, a weekly reporting agent, and the team version of the brain.
Here’s why marketers choose CXL
Keep up with AI native marketing tactics and workflows.
Subscribe to our weekly newsletter.
We find and report on AI native marketing case studies, success stories, and playbooks that are working out there right now.
Subscribe to our weekly educational newsletter