The last six months moved fast. Amazon launched an MCP server in February. OpenA
By Melissa Burdick · June 4, 2026 · Curated by George's Blog
The last six months moved fast.
Amazon launched an MCP server in February. OpenAI adopted the protocol in March. Salesforce, Google, Slack, HubSpot, and others quickly followed. More than a dozen major enterprise platforms went agent-accessible in a matter of weeks.
When AI tools started showing up in real workflows, people dove in immediately. Pull your data into the tool, ask your questions, get answers back.
It worked well enough that everyone wanted more. The next question became: why can't my AI just connect to this directly?
The old answer required a custom integration for every combination of AI tool and data platform. Every pairing, built and maintained separately. With so many tools and so many platforms in play, that model was never going to scale.
That's the problem MCP is solving.
The commerce version of this friction is clear. Your data lives across Amazon, Walmart, Instacart, and more. Getting AI to work with any of it meant manual steps every time. MCP changes that.
But the more interesting shift isn't just in the workflow itself. It's in what it means for the people and teams running it.
Riku Laitasalo, Pacvue's VP of Growth and Experience, puts it this way:
"It's the human plus AI that I'm really excited about. AI's not going to replace the human. It's going to make humans super managers or super stewards of commerce businesses."
Next week - I'll get contextual with use cases. And - comment below with yours! Let's compare notes.....