Dr. Irina Raicu.

Enterprise AI strategy · Operating models · Adoption

I turn AI strategy into operating capability.

Leadership teams come to me with working pilots and no system for deciding where AI belongs, who owns it, and what counts as value. I design that system and have run it at global scale.

  • Former Microsoft Global AI Director
  • Founder, Institute for Applied AI
  • PhD in AI

15+ years across consulting, AI product delivery, digital transformation, and organizational adoption.

Work with Irina See the enterprise practice
Dr. Irina Raicu photographed three-quarter to camera against a plain white background, wearing a navy double-breasted blazer with gold buttons.

Recognition

Where the work has been seen.

  • Microsoft
  • Meta
  • Tommy Hilfiger
  • Forbes
  • Garmin
  • Jing Daily
  • Harper's Bazaar Art
  • The Recursive
  • Prima TV
  • Antena
  • RFI
  • La French Tech
  • Institut Francais
  • NFT Bucharest
  • BIZ
  • Decentraland
  • DRESSX

Proof

What the work produced.

Selected outcomes from Microsoft enterprise AI adoption work

  • 90%+daily active use reached within nine months
  • 2,000+person global support organization across the Americas, EMEA, and APAC
  • Contributed to $3M+in documented savings, 2024
  • 4.1 → 4.6CSAT improvement, 4.1 to 4.6

Point of view

Most AI programs stall after the pilot, and the reason is rarely the model.

By the time a leadership team calls me, the technology works. What is missing is scope, ownership, and measurement that survives the first quarter.

That is an operating-model problem, not a data science problem.

AI does not replace the decision. It moves it to which work you hand a model, and how you check the result.

How I work with leadership teams

What I do

Four decisions I help leadership teams get right.

Engagements run as executive advisory, operating-model and roadmap work, leadership workshops, or executive education.

  • 01

    AI Operating Models

    At Microsoft I designed a hybrid hub-and-spoke model linking a central AI Innovation Hub to regional enablement leads, managers, practitioners, and a 65-person champion network.

    Typically includes

    • Operating model and decision rights
    • Central, regional, and business-unit roles
    • Intake, review, and funding cadence

  • 02

    Adoption & Workforce Redesign

    Adoption is a redesign of work, not a training plan.

    Typically includes

    • Role and workflow redesign around human-agent collaboration
    • Champion and enablement network design
    • Usage, quality, and behavior instrumentation

  • 03

    Use-Case Portfolios & Value

    I build ranked portfolios, not wish lists: 10+ use cases assessed; four advanced to production.

    Typically includes

    • Use-case intake, scoring, and sequencing
    • Value cases with baselines agreed before build
    • Post-launch measurement and portfolio review

  • 04

    Executive Roadmaps & Programs

    I translate between executive priorities, business workflows, and technical delivery, where programs break.

    Typically includes

    • Roadmap, dependencies, and decision gates
    • Board and executive committee reporting
    • Executive education and leadership workshops

See how engagements are structured

Selected enterprise impact

Programs I designed, ran, and measured.

Global support organization

Copilot and agentic AI adoption at global scale

As Microsoft Global AI Director, Enterprise AI Adoption & Support Transformation, I led Copilot and agentic AI adoption across a 2,000+ person global support organization spanning the Americas, EMEA, and APAC.

  • Reached 90%+ daily active use within nine months
  • Improved CSAT from 4.1 to 4.6
  • Reduced complex case resolution from 7+ days to approximately three days
  • Contributed to $3M+ in documented savings in 2024
Dr. Irina Raicu on stage at an Inspiring Women in Business event, standing beside a man handing a framed digital artwork down from the stage, with the event title slide projected behind them.

AI platform and delivery

Shipping models faster, and defending them

A portfolio is only credible if models reach production and stay supported.

  • Reduced MLOps model-deployment cycles from 3-6 months to under six weeks across 10+ production models
  • Improved threat-detection response by 90%
  • Contributed to an estimated 80% reduction in incident-response time

Sustainalytics / Morningstar

Automating research operations without losing the analysts

I redesigned document-heavy research work with the analysts who had to trust it.

  • Reduced manual document processing by 70%
  • Reduced report turnaround by 40%
  • Reached 90% analyst adoption

Société Générale

Building the center that makes AI repeatable

Co-founded an AI Center of Excellence and led transformation of a cloud financial platform serving 2M+ daily users and 500M+ euros daily, under regulatory review.

Earlier career spans Accenture and IBM. Research foundation: a PhD in AI, deep learning in the financial-banking domain.

Read the full enterprise practice

Speaking and executive education

I speak to the people who have to make the decision.

Sessions run as keynotes, executive workshops, or leadership sessions with the delivery team, built on numbers from programs I ran.

Recent and past audiences

Microsoft Partner EMEA · Cannes Lions · Meta Creative Days · Garmin · Harvard · University of Central Florida · London College of Fashion · Milano Fashion Institute · Antena Academy

Dr. Irina Raicu seated at a studio microphone during a REWIRED Podcast recording, with the episode quote set across the lower half of the frame. Play
70% of what makes AI work has nothing to do with the tools.
Dr. Irina Raicu on the REWIRED Podcast, a long-form conversation on adoption and operating models.

Watch the episode

Talks, video, and booking

Parallel practice

A separate body of work, and the reason it matters here.

My enterprise work designs how organizations operate with intelligent systems. My creative practice asks where synthetic imagery earns or loses trust.

A humanoid robot in an orange denim suit with a large round padded collar stands on stone opera-house steps beside a woman in a pale green pleated coat with lime feather trim, a long printed scarf trailing down the steps between them.

Robot Image Architecture

Non-human public figures built as characters, with an image system, styling rules, and a campaign world they hold over time.

Enter the practice
A woman in sunglasses and a robot mannequin with a lacquered black head, both in matching lilac denim printed with photographic panels, walking a red carpet beneath a gilded staircase.

Fashion

Couture, computed. Garment and editorial work that started in my mother's atelier and now runs through generative systems.

See the fashion work
A frontal portrait of a porcelain-white figure with pale blue eye makeup, a mechanical headdress, and a crown of pink and white blossoms, against a blue background.

Art

Image work on climate, memory, and how machine vision reads the world. Exhibited in the CVPR Art Gallery, 2024.

See the art

Creative practice overview

Work with Irina

Tell me where the program is stuck.

Useful first conversations start with a blockage: a portfolio nobody can prioritize, an adoption number that stopped moving, or a board that wants value evidence. Send a note. I will tell you if I can help.

Executive advisory · Operating-model design · Use-case portfolios · Leadership workshops · Executive education

Start a conversation View speaking