In the span of about three years, I've gone from using AI for point solutions (Jasper for writing, Midjourney for images.) to using it as the central nervous system for much of our marketing. For practically everything except layout and design. AI still falls short there.
For the longest time, I was skeptical of letting AI touch anything that really mattered. Now we run much of BrownRobinson through Claude and barely open most of the tools it's connected to.
Here's what changed: Claude earned our trust. And Claude can use MCP servers.
Without getting too techie, an MCP server lets Claude connect to and work directly inside the tools we (and you) already use: your CMS, marketing automation platform, project management system, internal knowledge base, client platforms, and more. Nothing gets copied and pasted between ten browser tabs, or exported as a CSV from one SaaS platform just to get re-imported into another.
Early MCP implementations mostly read data. Today they're creating, updating, and managing work across connected systems from within Claude.
Software still matters. But institutional knowledge matters more.
That's why we named our MCP "BR Pulse." BR Pulse isn't valuable because it's connected to Webflow, HubSpot, Notion, or dozens of other applications. It's valuable because twenty years of our frameworks, positioning models, GTM thinking, ICP development, messaging principles, competitive analysis, and marketing playbooks are built into it.
The magic isn't that Claude has access to our tools. It's that it has access to how we think.
We give Claude the right client context, market intelligence, and frameworks. From there it performs research, competitive analysis, messaging development, sales enablement, customer marketing, and other work with context instead of guesswork.
Work that used to take us a week now takes about thirty minutes. Not because AI is smarter than marketers. Because it never gets tired of the tedious parts.
We still review everything ourselves for consistency, formatting, factual accuracy, and judgment. And like any good software, BR Pulse is constantly evolving as we refine its skills, frameworks, and workflows.
Here's an example of where it really clicked for me: Customer marketing has always been a big part of my playbook. Cross-sell and upsell are often the most overlooked growth opportunities in a business. Unfortunately, onboarding emails are usually the forgotten part of marketing. Product names change. Features evolve. Messaging shifts. The emails often don't, or aren't top priority.
Before long those emails are super generic, outdated, and largely ignored. I've seen campaigns still describing capabilities as "coming soon" months after Sales had already been selling them.
A few years ago, updating those campaigns meant opening every email in HubSpot or Marketo, cross-referencing product notes, rewriting every message, testing everything, publishing it, and then hoping someone remembered to revisit it six months later.
Today I'd give Claude the product update, access to the campaign, and a clear definition of what "done" looks like: present tense, no early-access language, updated FAQs, consistent messaging, and personalization based on what Sales learned during the buying process. The same standards and outputs I'd expect from a strong product or customer marketing manager.
A few approvals and clarifying questions later, fifteen emails across different audiences and implementation stages were rewritten, along with a summary of every meaningful change.
My job on that project became review and fine-tuning, not authorship.
We're not replacing marketers. We're eliminating the work marketers hate. The biggest shift wasn't technical. It was trust.
BR Pulse follows the guidance we wrote every single time. Tone. Structure. Positioning. Completeness. ICP alignment. Looking for adjacent buying influences and opportunities we might otherwise miss.
That's why we trust it. AI isn't another app I use. Once it's trained, it's another teammate.
Marketing is already ending up being one of the biggest beneficiaries of this shift. Updating CMS content. Revising campaigns and automations. Cleaning and better contextualizing CRM data. Competitive research. Reporting. Content operations.
The work gets dramatically faster once AI can operate inside your systems instead of outside them. Most companies are still treating AI like a chatbot: Rewrite this. Summarize this report. Make me a slide from this spreadsheet.
The real shift happens when AI can work inside your business. That's where the time savings show up. Not from asking better questions, but from eliminating the manual work between systems.
Building an MCP isn't nearly as technical as most people assume. The hard part isn't the code. It's documenting how your company actually thinks and works. Your frameworks. Your standards. Your workflows. Your judgment. That's what you're really programming.
But this synthesis -- these outputs -- this finally feels like AI delivering on its promise. Not another shiny object. Real deliverables. Better workflows.
I wouldn't have trusted this a year ago. I wouldn't have trust it six months ago. Now it's just how we work.
