Dear reader, as a quick introduction: I am a Salesforce Marketing Cloud consultant at Merkle. This is why most of my content is oriented around Data Cloud, Agentforce, and SFMC.
In such a fast-paced environment, where the volume of news and updates is constant, it can be hard to keep up with everything that is happening. New features, new releases, new best practices, every week.
I decided to automate the whole thing. I built a multi-agent AI system that runs every week, collects the latest content from the sources I care about, evaluates it, and delivers a structured report straight into my Notion workspace.
One button. One click. A few minutes later, the report is there.
This article walks through how I built it, the architecture behind it, and the decisions I had to make along the way.
There are too many sources to follow manually. Official documentation, blogs, community forums, influencer content.
Every week, something new comes out. And as a consultant, not knowing about a feature that your client asks about is not a good look.
I wanted a system that collects everything, filters the noise, and gives me a clean summary. That is what led me to CrewAI.
CrewAI is an open-source Python framework for orchestrating AI agents. You define agents, give them roles and tools, and they work together on a task.
Four concepts to know:
The framework handles the orchestration. Your job is to define the right agents with the right prompts and the right tools. Simple to understand, but getting it to work reliably in production takes real iteration.