Your degree is the foundation.Keystone builds the bridge.

Agents read your syllabus, real job ads and your public work, then plan only what your degree doesn't teach yet, with a pod of students to build it with.

Aisha's dashboard, Year 1, BSc CS

38% ready for ML engineer roles

84% by Sep 2027 if she follows her roadmap.

This week: Docker

GapRising

41%of ML engineer ads ask for Docker, and none of her modules teach it.

The demo follows Aisha, a first-year Computer Science student at Manchester who wants to become an ML engineer.Demo data

Learning, assessment and teaching, joined up around the degree you already have

  • Learning

    A week-by-week roadmap that skips what your modules already teach, with a Why? receipt on every step.

  • Assessment

    Skills count when there's proof: a merged PR, a micro-challenge or a mock interview. Never a self-rating.

  • Teaching

    You teach a pod-mate what you know. Lecturers see anonymous cohort gaps and lab ideas for groups of 10 or more.

How it works

Four steps, each one a page in the demo. Every step has a Why? so you can see the numbers behind it.

  1. Bridge Map

    Your syllabus, real job ads and your GitHub on one map. For Aisha: 4 skills proven, 11 covered later by her degree, Docker the biggest gap.

    See the Bridge Map
    Aisha's Bridge Map in Keystone: a skills graph in four columns, proven, in your degree, later in your degree and gap, with Docker selected at the top of her 13 gaps
  2. Roadmap

    Weeks that build on your modules, skip what they'll teach you anyway, and go lighter in exam weeks.

    See the Roadmap
    Aisha's roadmap in Keystone: week 1 is Docker, with a learn, practise and build task and her proof for the week, then week 2, PyTorch
    Skipped: Statistics, Linear algebra, Databases and 8 more. Her degree teaches them later.
  3. Pod

    Two or three students who are strong where you're stuck. Aisha learns Docker from Priya and reviews Mei's Python.

    See the Pod
    Aisha's pod project in Keystone: VisionAid, an accessible image captioning API, planned as a 3-week project with one milestone a week
    Pod Orbit: Priya reviews her Docker PR, and she reviews Mei's Python.
  4. Proof

    A real GitHub project. When a teammate approves your PR, the skill moves up your Skill Passport.

    See the Skill Passport
    Aisha's Skill Passport in Keystone: Python, Git and Data structures each sit on the Tried it rung, with Built it and Can teach it still to climb, and every rung links to its proof

Proof climbs a ladder

Keystone turns what you're taught into what you can prove. Each skill on your passport sits on one rung, and only real work moves it up.

  1. Tried it

    You finished a micro-challenge

  2. Built it

    A pod-mate approved your merged PR

  3. Can teach it

    You reviewed someone else's work in it

Questions students ask first

Does Keystone grade me?
No. Agents propose and you decide. A skill counts only when there's proof you chose to share, and every recommendation explains where it came from.
What does it read?
Your programme's public syllabus, job ads, and public work you link, like your GitHub. Nothing is posted anywhere on your behalf.
How are pods matched?
On skills, interests and availability. Never on personal characteristics. Badges like New to UoM are optional and only send help your way.

Built for Reply's AI on Campus

Specialist agents hand work to each other, one knowledge graph joins modules, job ads and people, and Bridge Week takes it to campus.