I counted the clicks. From the LinkedIn home feed in Dubai, on a fresh logged-in account, to a submitted application for a senior product role at a UAE-based AI company, the path is eleven clicks across four distinct UI surfaces. Five of those clicks are dismissals — modal closures, "not now" buttons, and back-arrows out of pages I never asked to enter. Two of them are corrections — re-applying the country filter after LinkedIn's recommendation engine quietly replaced it with a US-default. The remaining four are the actual application flow. The signal-to-noise ratio of the LinkedIn jobs surface, for an actual job seeker in an actual country trying to apply to an actual role, is roughly 36 percent. Two thirds of the work the user does on LinkedIn's jobs UI in 2026 is fighting LinkedIn.
I'm not writing this to dunk on LinkedIn. They're the dominant professional network in the world, they have a defensible business, and they have shaped recruiter workflow for two decades. I'm writing this because the gap between what LinkedIn is good at — networking, signalling, recruiter sourcing — and what it has become for the job-seeker — a confused hybrid of social feed, learning storefront, premium upsell, and reluctant job board — has widened enough that we shipped an entirely different product to fix it.
ENTRA Careers, the platform side of what I'm building, was built on a single design constraint: a job seeker in a specific country, looking for a specific role, in a specific industry, should reach a list of qualifying jobs in three clicks or fewer. Country, industry, role. Three filters, set once, persistent across the session, never quietly overridden by a recommendation engine. Geography is the primary axis. Everything else is secondary. The product will not show you a US role when you have set your country to UAE, regardless of how many of LinkedIn's product managers' KPIs depend on cross-pollinating the feed.
The structural problem with LinkedIn's jobs UI in 2026 is that the surface has been optimised over twenty years against four different objectives that no longer compose. Recruiter-side discovery wants the largest possible candidate pool, which pushes the geography filter to be soft. Premium upsell wants the candidate to feel friction, which pushes more modal interruptions. Feed engagement wants the user to scroll, which pushes the jobs UI to surface adjacent content (learning, news, posts) at every step. Application conversion is the fourth objective and is structurally the lowest priority of the four. The result is a product that does many things adequately and none of them outstandingly.
The recruiter-side conversation here is more nuanced than the founder-tweet caricature usually allows. Recruiters at every Fortune 500 company we've spoken to in the last quarter — and at every frontier-AI lab, where the senior-IC market is so contested that funnel mechanics matter even more — have privately confirmed that LinkedIn's 2025 UI changes broke long-standing sourcing workflows. Boolean search degraded. The recruiter inbox migrated to a chat-style threading that buried longer outreach. Saved-search persistence got less reliable. The recruiter community migrated, partially, to GoodTime, Gem, and a small handful of LinkedIn-alternative sourcing tools. The candidate side of the equation has had no equivalent migration target — until now, it has been LinkedIn or nothing.
Building a job board in 2026 is, by any reasonable analysis, a bad idea. The category is mature, the incumbents have distribution and brand, and the unit economics of pay-per-post job boards have been compressing for ten years. We did it anyway because the gap we observed in the candidate-side product is structurally different from the gap the incumbents are competing for. We are not trying to be a better LinkedIn. We are trying to be the geography-first counterweight to LinkedIn — the place an AI engineer in Dubai, a fintech product lead in Singapore, or a biotech researcher in London goes when they want to see the jobs in their country, in their industry, ranked by quality, without fighting a feed.
Three design decisions we made that LinkedIn structurally cannot make.
The first: country is a hard filter, not a recommendation. Setting your country in ENTRA does not surface "jobs you might be interested in elsewhere." It surfaces the jobs in your country. Every time. The recommendation engine works inside the country filter, not around it.
The second: the resume upload is gone. I wrote about this separately in the resume-upload founder letter. The short version: a resume upload is friction without proportional signal value, and we removed it on day three of the product's life. A Verified Profile — a structured, edit-once career-shape page — replaces it. The conversion lift was 34 percent.
The third: there is no social layer. No feed. No "people you may know." No congratulations-on-the-new-job posts. The product is a job board. It does the job-board job. It does not try to be three other products at the same time.
The forecast: LinkedIn will fix some of this. The 2025-vintage UI changes that broke recruiter workflow are already being walked back in beta, and the company has the engineering depth to rebuild the candidate-side surface if it decides the gap is worth closing. What LinkedIn cannot fix is the structural conflict between the four objectives the surface is being asked to balance. As long as feed engagement and premium upsell are co-equal with application conversion, the candidate experience will remain the lowest-priority workload on the surface. That is the gap a focused product can hold.
For the broader picture of how Fortune 500 hiring is being rebuilt around the failure modes of 2024-vintage tooling, see Why 17 Fortune 500 CHROs killed their resume-screening AI. For ENTRA's own funnel philosophy at the senior-IC level, see How Anthropic restructured its talent stack.