LinkedIn Newsfeed Agent
A product concept and interactive case study for LinkedIn's Senior PM (Feed) role — a Newsfeed Agent that activates on Save and turns passive engagement into completed professional actions.
The problem
LinkedIn's feed drives massive engagement — comments +37% YoY, video watch time +36% YoY — but most of that intent evaporates. Save is the one signal that captures real professional intent, and it sits idle. You save a job post, a thought leader, an article, and nothing happens next.
My hypothesis
If I could pitch the Senior PM (Feed) role with a real product concept instead of a cover letter describing one, the case study itself would show the PM thinking. The bet: a scrollable, interactive web app grounded in LinkedIn's own earnings data is more persuasive than any doc.
What I built
A scrollable case study at linkedin-feed.schlacter.me framed around one product concept — a Newsfeed Agent that activates on Save. The page opens with LinkedIn's real momentum metrics (pulled from Q2/Q3 FY25 earnings), then builds the argument for why Save is the strongest intent signal and what an agent layer on top of it unlocks. The interactive prototype lets you pick a scenario — Job opportunity, Someone to reach out to, Article to act on — and see the agent draft the completed action in context: 'Apply for me,' 'Draft the outreach,' 'Summarize and ask what you want to do with it.' The Premium upsell is designed-in: agent actions are gated above the free tier, extending the $2B+ Premium flywheel.
What broke
The hardest part wasn't the prototype — it was writing the case study without sounding like an outsider telling LinkedIn what to build. I rewrote the framing three times to land on opportunity-only language grounded in LinkedIn's own earnings. The demo surface is a mockup, not a working agent: I'm simulating the intent-capture flow, not running a real model on real saves.
What I learned
A product concept grounded in a company's own public data reads differently than one built on outside assumptions. Anchoring to LinkedIn's earnings numbers (+37% comments, $2B+ Premium run rate) did more work than any framework I could invent. The other lesson: for a feed product, the demo has to feel like the feed — a deck or doc couldn't carry this.
If I kept going
If I get the interview, build a second scenario — the Silent Sleuther mode, where the agent surfaces career signals without requiring a visible engagement. Also add a measurement section: save-to-action conversion rate as the new leading indicator, with a napkin model of what moving it from ~0% to 20% would unlock for Premium and Jobs revenue.
Prototype
The thesis: LinkedIn's feed drives massive engagement — AI can make every interaction count. Earnings data from Q2/Q3 FY25.
The save signal: real Reddit quotes showing members treating Save as an intent signal LinkedIn hasn't yet acted on.
The prototype: pick a scenario — Job opportunity, Someone to reach out to, Article to act on — and the Newsfeed Agent drafts the completed action.