Universities Track Graduation Rates. Nobody Tracks Whether Students Get Jobs.

01 /The Concept

Getting a job after graduation sounds simple until you watch most students try. CareerOS treats the job search as a structured, gamified process – built for university students navigating the gap between campus and career.

Three things drive the platform's effectiveness:

- Each student gets personalized recommendations based on their major, academic performance, and career preferences.

- University career staff get visibility into each student's activity on the platform – who's doing the work, who's coasting – so they can intervene before it's too late.

- The interface is designed to match how millennials and Gen Z actually engage with things, which means higher participation in the activities that lead to job offers.

The job search is framed as a game – a series of quests the student completes to accumulate points.

Each week has a target score. Points are earned for: saving cards of target companies, making new contacts with recruiters and industry professionals, and sending follow-up messages to maintain those relationships.

Default weekly targets are set by the platform, but high-activity students can push their own goals higher. The platform tracks scores in real time and displays a live leaderboard – injecting some competitive energy into what is otherwise a tedious process.

The professional networking layer works like a social feed: students see updates and posts from recruiters and experts they follow, can comment, and can subscribe to new contacts discovered in the feed.

The platform's core mechanic, the founders say, is notifications and nudges – "a recruiter just posted something relevant to you," "you have a new message," "you're dropping in the rankings." These keep students engaged and moving.

Beneath the gamification sits a serious AI engine that builds personalized job-search plans for each student, surfaces the most relevant opportunities, and helps draft outreach messages.

Founded in 2023 in Spain, CareerOS has already signed university clients in the US, France, and Germany. The startup recently raised its first round: $1.2 million.

02 /Why It Matters

The most successful startups help users reach their actual end goal – not just make a process easier, faster, or prettier.

The real end goal of a university education, for a student, is a good job. For the university, it turns out the goal is the same: the more graduates who land strong employment, the easier it is to recruit the next class. Employment outcomes are a far more compelling pitch than "quality of education."

CareerOS helps both parties hit their actual goal. This sets it apart from most edtech platforms, which optimize for the learning experience itself rather than for where that learning is supposed to lead.

One thing worth flagging: CareerOS markets itself as helping "the best students find their dream job." That framing is a bit off. The best students figure out job searching on their own – they're self-motivated and don't need nudging.

The real problem at most universities is the students who aren't actively looking. And at most institutions – a few elite programs aside – that's the majority.

Those students need consistent monitoring and encouragement to do anything at all. That monitoring costs the university staff time, and time costs money. CareerOS lets career offices scale that attention to far more students without burning out their advisors.

This is exactly the logic behind Siro, [covered last fall](/review/a-eshhjo-nuzhen-otlichnyj-plan). Siro built an AI sales coach – but the real insight isn't that it makes top reps better. Sales performance in most teams follows a steep exponential: a small percentage closes most of the business, and managers naturally focus their coaching time on the stars.

The underperformers are largely ignored. But many of them can improve with sustained coaching – which human managers won't provide because their time is too scarce. AI can. The goal is to flatten the performance curve so the bottom half of the team stops dragging down overall results.

This connects to something broader about deploying AI inside organizations. Most companies instinctively start AI adoption with their best employees – but that's precisely the wrong move. Research shows strong employees improve by about 17% with AI assistance; average employees improve by 43%, narrowing the performance gap between the two groups from 22% to roughly 4%. The ROI on AI-training weaker performers is several times higher than investing in your stars.

The practical implication: when you scale, you hire in volume – which means you hire average people. Making average people more effective is the only way to hit your numbers at scale. AI is how you do that. And since learning to coach underperformers with AI takes time to dial in, starting earlier is a compounding advantage.

03 /Opportunities

A broader lesson for any startup embedding AI: build for the people who are bad at the thing, not the people who are already good at it.

AI can spend unlimited time coaching, explaining, and doing the work alongside underperformers until they reach an acceptable baseline. That's a dramatic improvement for them – and the kind of improvement they or their organization will pay for.

The question to ask: what are large numbers of people currently doing poorly? Could they do it significantly better with substantially more support? Can AI provide that support? What would it need to know and do?

A more specific angle is CareerOS's domain itself.

CareerOS targets universities – but the same problem exists across the online professional training space. Bootcamps and certification programs market themselves on employment outcomes. Low placement rates are rarely a teaching quality problem; they're a job-search-effort problem. Students don't hustle enough during the program, and instructors can't afford to individually guide every student through the application process.

An analogous platform targeting online course providers – rather than universities – could work well, with two additions:

- A built-in library of job search training content: networking scripts, resume writing, interview prep.

- Centralized employer partnerships, giving hiring managers a single funnel into graduates across many programs – valuable to employers who are actively struggling to find qualified candidates.

The opening here is real: most edtech focuses on improving learning. Almost no one focuses on improving the outcome that learning is supposed to produce. That's the gap.