Applied AI Manager / AI-native transformation · Dubai

I would run Careem's AI-native transformation the way Careem builds product: inside the workflow, shipping.

I am a transformation operator who does not just scope AI programs, I build and ship them myself. Ten-plus years of operating-model and venture change, and AI products I have taken end to end. I am moving to Dubai to do exactly this work.

Straight up: the Applied AI Manager req has closed. I built this anyway, because it is the actual week-two work, and because WorkOS says Careem is not slowing this down. Treat it as me raising my hand for where this goes next.

Open the Redesign Desk See how I operate

Swiss Post

AI Project Lead, turning AI ideas into go-to-market and new revenue. Since Jan 2026.

Pedal Peak

A live product I built end to end with AI tooling. Real users, in production.

Zero to one

Founding team, two rounds raised, scaled to ten locations. I ship, then scale.

// the centrepiece

The Redesign Desk

The job names one judgment above all others: for a given workflow, do you augment it (AI-enabled, keep the process, add a copilot) or rewrite it (AI-native, zero-base it around agents)? Pick a real Careem workflow, make the call, and watch the system, the stack, the ROI, the rollout and the guardrails change with it. This is the meeting we would have in week two.

reading the numbers on the desk careem a real, sourced Careem number · illustrative my estimate, set without inside data, rebaselined in week two

posture: augment

CARE · CUSTOMER SUPPORT

Captain support at 50k tickets a day

the bottleneck

Make the call for this workflow:

Ramona's call

// the operator behind the desk

Every claim on that desk has a receipt.

The desk works because I have done the parts it is made of: reading a business for where the value is, building the thing, and getting people to actually use it. Left is the claim, right is the engagement that backs it.

I turn AI ideas into revenue, not slideware.

At Swiss Post, as AI Project Lead I size the opportunity, build the prioritised roadmap with KPIs, make the build-vs-buy and cost-vs-benefit calls, and carry it from concept to launch.

I build with AI myself. Hands on keyboard.

Pedal Peak is a live cycling product I built end to end with AI tooling (SvelteKit, Cloudflare, Strava API, n8n). Real users. This is the "builder/operator" the role asks for, evidenced.

Zero to one, then scale with rigour.

Founding team at WePractice: two funding rounds closed, grown to ten locations and twenty-three people, a thousand-plus client matches in year one. I have done the "then scale with observability" part.

I move operating-model change at P&L scale.

At Ifolor I owned the e-commerce operating model for a CHF 100M-plus business and lifted conversion nine percent and checkout step-rate fifteen percent through research, testing and analytics.

Systems thinking is where I started.

Accounting to FMCG product marketing to corporate innovation (Mobiliar: MVPs to launch, Design Thinking) to venture building. Mapping a messy process to the intervention that moves it is the through-line of the whole arc.

// the honest read

Three things you would rightly want to pin down.

I would rather name the gaps than have you find them. Here they are, with the counterweight.

Your AI title is recent.

True. AI Project Lead since January 2026. But the building is not recent: Pedal Peak has been live and AI-built since 2023. The title caught up to the practice, not the other way around.

Are you an ML engineer?

No, and I would not claim it. I select models, engineer prompts, and ship LLM tools and automations myself, then partner with AI and software engineers on the deep pipelines. That is exactly the split the role describes.

You are not from consulting.

Correct. My transformation track record is from running the change, in ventures, in e-commerce, in innovation, not advising on it from outside. For a builder/operator role I read that as the right side of the line.

// the medium is the message

This is a work sample, not a brochure.

I designed and built this whole page, and the logic inside the Redesign Desk, with the same AI toolchain I would bring to Careem: Claude Code plus my own prompt and automation stack. The desk is not a slide of a demo. It runs real logic, recomputes real trade-offs, and cites its sources.

If you want to know how I would work on your Care queue or your Pay disputes, you are already looking at it. This took an afternoon. Imagine it pointed at your real data.

// the ask

Careem is putting agents into Care, Pay, Food and People. I want to be the operator who ships them.

The role I clicked has closed, so this is me reaching out directly. If WorkOS or the AI-native transformation work has room for someone who makes the augment-versus-rewrite call and then builds it, let's talk. Fifteen minutes is plenty to start.

Ramona Furter

AI Project Lead · transformation operator · moving to Dubai

ramona.furter@icloud.com +41 78 907 14 47 linkedin.com/in/ramona-furter

scan for LinkedIn

Read the full CV