Back
  • 2026
  • Visionary work
  • B2C + B2B
  • AI strategy
  • XING
  • 20M+ members

From AI strategy to a 0→1 hiring vision

My role

I led the design translation of an AI product strategy into a concrete hiring journey vision. My concept helped shape the direction in an SVP-led workshop, then became future B2C and B2B flows, animated vision artefacts, and fast-tracked experiments.

Opportunity

Finding and filling jobs is slow and effortful — for both users and recruiters. Research showed a clear need for time savings, reduced effort, and better matching. The opportunity was to rethink the hiring journey end-to-end: using AI to take over the heavy lifting while improving efficiency, match quality, and application quality.

Outcome and Impact

Turned abstract AI strategy into shared product direction — from leadership vision to tangible flows, implementation specs, and first experiments. The initial shadow-profile test validated the approach as a safe foundation, with no drop in engagement.

Miro board with early AI hiring journey concepts and research insights
To the drawing board

Two days of focused vision work with leaders from engineering, product, design, and marketing, facilitated by the SVP of Product. I helped translate fresh user-research insights into concrete product directions — focusing on where AI could meaningfully reduce effort across the hiring journey.

AI that moves from intent to action

These flows explored how an assistant could ask just enough, infer useful context, show tailored results and help users take the next step — while keeping the user in control. The wider vision was tested with 598 jobseekers: 86% liked the overall idea, and 80% liked applying in chat.

AI search from two questions

Two simple prompts become a profile-informed job result stack. The assistant gathers intent, infers context in the backend and turns a short conversation into useful results.

From profile to application

An AI assistant prepares a role-specific CV from the user’s email and LinkedIn profile, while keeping final review and submission in the user’s hands.

A whole heap of vision — made watchable

To make an abstract vision tangible, I created a series of animated screens — a shared artefact that helped leadership align, teams converge, and roadmap decisions take shape.

Getting real — implementation

I kept design and engineering aligned from concept into build — defining specs, resolving edge cases, and collaborating daily to ship the fastest viable version of the shadow-profile builder.

Normal-flow specification for entering, editing and clearing two input fields.
Flow specification for the first-field-only path and adding the second input.
Normal-flow specification for entering, editing and clearing two input fields.
Specs don’t stop at visuals — they define behaviour.

Even two input fields require decisions: when values persist, how clearing behaves, and when additional inputs appear. These rules are not edge cases — they are the product.

Dark mode version of the AI job search entry screen
Pssst...for the people on the dark side of the moon

Dark mode changes how interfaces communicate hierarchy, contrast and depth. Instead of adapting designs later, we built it into the system from the start — defining both modes in the specs to ensure consistency and clarity.

Annotated discarded screen directions with red crosses and one selected direction
Show where you didn’t go

Instead of hiding discarded directions, we documented them. This made decisions easier to explain, prevented repeated loops, and helped the team align faster around the path forward.

Two mobile screens showing height-aware job card layouts
Responsive design isn’t just width — it’s height.

For swipeable cards, height introduces a real trade-off: fixed layouts waste space, flexible ones lose content. We made cards height-aware — maximising content while defining clear prioritisation rules.🤓

Height-adjustable job card prototype showing content truncation, job details and skill tags.
Interactive! Prototype the behaviour, not just the screen

I built a CodePen prototype to define truncation logic, test edge cases, and align design and engineering on consistent behaviour across screen heights.