The Problem
PressW’s team of AI experts share tons of awesome new resources on the cutting edge developments in our field with on another on a weekly basis. AI moves fast, and we need to keep up to continue providing the best in class to our clients.
These articles and best practices are shared across a variety of slack channels, notion pages, and more on a weekly basis. Not only this, links range from sources such as X.com to YouTube videos, to medium articles, to scientific publications. Plus, each new resource often generates a lot of discussion amongst the team members who see it. How does a PressW employee keep up with so much disjointed information?
Enter PressW’s newest AI creation — The AutoNewsletter. We now have a system that aggregates information across different internal channels, parses from various external sources, and summarizes in a single place, delivered directly to each employee every Monday morning.
The Brief
We’ve created a smart internal tool that helps us stay on top of all the great information our team shares. It’s like having a really organized librarian who never sleeps! Every week, our tool gets to work, gathering all the interesting links and insights our team has been sharing on Slack, Notion, and other platforms. But it doesn’t stop there — it’s pretty clever about how it handles different types of content. For example, when it comes across a post from X.com, it knows how to extract the key points. If it’s a YouTube video, it can pull out the main ideas. It even knows how to handle audio clips, regular articles, and PDFs. Each type of content gets its own special treatment to make sure we’re not missing any important details. This approach helps us make the most of our collective knowledge. Instead of valuable insights getting lost in the shuffle of busy workdays, our tool makes sure they’re captured and easily accessible. It’s not just about collecting information — it’s about turning that information into a resource that helps us learn, grow, and come up with better ideas for our clients.
Once ingested, we hand off the info to an LLM (Large Language Model) agent that summarizes each article with an eye toward brevity and applicability to our team. Now, the system uses another LLM to format the outputs in an easy to read Markdown format and places the output in an email template to be distributed via SendGrid right to each member of our team!
The Bottom Line
- Saved ~4 hours per employee per week of reading + manually sifting through topics that may or may not apply
- Supported continued education on hot AI/ML topics that otherwise employees weren’t finding or were missing
