Building our full-stack AI marketer in public. This is step one.
Our Journey: Building Claude into a Full-Stack Marketer with MCP
By Rodrigo Pozo · 2026-05-15
Category: Marketing
We are a small team. We have to be deliberate about every tool we add and every decision we make. This is the first in a series where we share exactly how we are building our marketing stack using Claude as the AI brain.
The belief we are starting from
Marketing has always been the art of going to market. Not a science. Not a funnel. An art.
The marketers who build enduring brands are not the best optimisers. They are the best interpreters. They read culture before it trends. They understand that a price is not just a number but a signal. That a campaign is not just a message but a relationship.
AI does not replace that craft. It makes it accessible to more people.
For a long time the gap between having a good marketing instinct and acting on it was enormous. The big players had research teams, brand tracking studies, and agencies. Everyone else had gut feel and guesswork. That asymmetry shaped which brands got built and which stayed small.
That gap is closing. Not because AI replaces the judgment. Because it removes the friction between having an instinct and validating it.
The architecture: two jobs, one connected system
A full-stack marketing setup needs two things before anything else.
A strategist and orchestrator that holds the 4 Ps together, keeps strategy grounded in how growth actually works, and makes sure the work connects to the fundamentals rather than chasing whatever tactic is trending.
Ongoing market intelligence — a continuous read on what is happening in your category, where demand is growing, what your audience is actually searching for, and what competitors are doing.
For both we use MyTelescope as an MCP inside Claude. A core skill trained on the marketing fundamentals: the 95/5 rule, 60/40 thinking, how pricing signals brand value, how to build emotional communication across the full funnel.
From that core we have broken the work into specific skills, each built for a different job. Market diagnosis. Brand tracking. Audience insights. Competitor intelligence. Campaign briefs. They are designed to cross-reference each other so when Claude works on one job it can pull context from the others. The whole is smarter than the parts.
The tools we are using today and why
MyTelescope via MCP — Our orchestrator. Demand signals across 11 channels. The market context that underpins every other decision.
Windsor.ai — Aggregates paid media from Google, Meta, and LinkedIn into one layer Claude can query. One conversation instead of three dashboards.
Fibler — Our content analytics. Connects the market picture to our own performance.
Prisync — Competitor pricing connected to demand signals. Tells us whether volume changes are price-driven or demand-driven. Different problems, different solutions.
Google Analytics 4 — Where demand lands and what it does when it gets there.
HubSpot — CRM and pipeline. Where market demand connects to revenue.
Canva via MCP — Once Claude writes a brief grounded in real signals, Canva builds the asset. Brief-to-visual in one workflow.
Google Drive — Strategy docs and briefs. Claude reads and writes into our existing workspace.
Slack — Monthly demand summaries and alerts posted directly to team channels. Intelligence arrives where the team already is.
Claude Code and Skills — Where the architecture lives. This is what makes Claude domain-specific rather than general. Without skills, Claude is smart. With them, it knows the category, the brand, the competitive context, and the marketing fundamentals.
What the output looks like each month
A smart market dashboard saved to MyTelescope, updated from live signals, interrogatable in plain language. A monthly signal report covering category trends, brand shifts, competitor moves, and pricing windows. Campaign briefs grounded in what the audience is actually searching for, not the language we prefer internally.
The honest assessment
This is not a finished system. It is a working one.
The biggest shift is not that AI does more work. It is that work which used to require switching between twelve tools and manually reconciling outputs now happens in one conversation. The market context, ad data, pricing signals, and CRM picture are all available to Claude simultaneously.
A small team with this setup has the same quality of market intelligence that a large marketing department had three years ago. That is the democratisation. Not of the job. Of the craft.
More to come.
Follow this series on the MyTelescope blog and on LinkedIn.
Ready to build your own full-stack AI marketing setup? Connect MyTelescope to your AI →