CORE Intelligence.
I built an AI intelligence pipeline that turned hours of manual research into a five-minute daily briefing for Compass Group leadership.
While at Compass Group's CORE Solutions division, I noticed our leadership team was spending too much time sorting through industry news from dozens of different sources. I built an AI-powered intelligence pipeline that automatically filtered more than 25 RSS feeds against our strategic priorities and delivered only the most relevant technology updates directly into Microsoft Teams. What used to take hours each week became a five-minute daily review.
Ingested Daily
Saved Weekly
Using It
Built It
Data flows one direction: feeds are ingested, classified against Compass's strategic priorities, and delivered to executive leadership as a daily briefing.
From manual scan to automated signal.
Three phases from problem to production, built end-to-end in a matter of weeks.
CORE Solutions was Compass Group's internal innovation team. We focused on improving operations across supply chain, sustainability, food service, and enterprise technology by identifying better ways to solve problems before they became business challenges.
Part of that work meant staying ahead of where technology was going. New AI tools, automation platforms, startups, and industry trends were emerging every week, and leadership needed to know what was actually worth paying attention to.
The problem wasn't a lack of information, it was too much of it.
Executives were spending hours every week reading industry publications, vendor announcements, newsletters, and trade articles trying to separate meaningful innovations from marketing noise. It was valuable work, but it took time away from strategy and execution.
I didn't want this to become another software purchase.
Instead, I built the entire solution utilzing platforms Compass already had in place. That meant no vendor evaluation, no implementation project, and no consulting engagement. Just a lightweight system that could be deployed quickly and improved over time.
The goal was simple: take manual filtering out of the process.
Instead of reading dozens of articles every day, leadership would only see the ones that aligned with CORE Solutions' strategic priorities, along with a short explanation of why each one mattered.
The system worked in three steps:
- Collect: Custom Python scripts gathered articles from more than 25 RSS feeds covering AI, supply chain, foodservice, and enterprise technology.
- Prioritize: Using Microsoft Power Automate and AI Builder, each article was evaluated against the priorities we'd defined. Relevant articles moved forward, while everything else was filtered out.
- Deliver: The highest-value articles were automatically posted into a Microsoft Teams channel with a short summary, giving leadership a quick daily briefing instead of another inbox to sort through.
I built the entire solution myself in just a few weeks. The hardest part wasn't writing the code, it was teaching the system what "strategically relevant" actually meant. Once that framework was in place, it could consistently identify the articles that deserved attention across very different types of industry news.
The first rollout went to the CORE Solutions leadership team. Since they had been doing the manual research themselves, their feedback helped fine-tune the system over the first few weeks.
As the recommendations improved, leadership naturally shifted from searching for information to acting on it. Instead of spending hours scanning industry news, they started each day with a curated Teams briefing; returning an estimated 5–10 executive hours every week.
Small system, disproportionate value.
- 5–10 hours per week of executive review time returned to strategic work and decisions.
- 3–4 executives actively using the system as their primary intake for emerging-technology signal.
- Zero incremental spend: built entirely on Microsoft tools Compass already licensed.
- Shifted leadership conversations from "what's out there" to "how do we act on this."
- Established a template for AI-native tooling: small, in-house, built to remove a specific friction.
The best AI tools aren't the biggest. They're the ones that remove a specific piece of friction from a specific person's day.